The Docket

California: 13 AI bills signed, including human review of automated discipline and AI provenance

Signed 13 bills. Within scope: SB 947 requires a person to review automated employment disciplinary decisions; AB 2713 and SB 1000 amend the California AI Transparency Act on provenance data for AI-generated material; SB 1159 adds AI transparency and governance requirements. Others cover workplace surveillance, health care, legal practice and digital replicas.

What it asks to be measured. Provenance data that lets AI-generated material be identified (AB 2713, SB 1000); human review of automated decisions in employment discipline (SB 947).

“AI should expand opportunity – not come at the expense of workers and families.”Gavin Newsom, Governor of California, official press release

Source: Office of the Governor of California. Checked Oct 08, 2026.

Limits: Bill-level detail is the Governor's one-line summaries; individual bill texts not retrieved.

Cite. Protocol STAMP, “California: 13 AI bills signed, including human review of automated discipline and AI provenance,” The Docket, entry us-ca-bills-2026-09-30, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-bills-2026-09-30.

US Executive Order 14434 replaces 'Artificial Intelligence' with 'Super Intelligence' in executive-branch documents

Executive Order 14434 directs agencies to use 'Super Intelligence' and 'SI' instead of 'Artificial Intelligence' and 'AI' in correspondence, communications, websites, reports and other non-statutory documents, and asks the Assistant to the President for Science and Technology to propose a federal definition. The President announced the name at the UN General Assembly on Sep 22, 2026.

What it asks to be measured. None stated. Until a new definition is adopted, the order defines the terms by the statutory definition of artificial intelligence in 15 U.S.C. 9401(3); definitions decide what evaluation rules cover.

“hereinafter officially called 'Super Intelligence.'”Donald J. Trump, President of the United States, at the UN General Assembly, official White House release

Source: The White House (Executive Order 14434) · The White House (fact sheet) · The White House (UN address). Checked Oct 08, 2026.

Limits: Federal Register citation not retrieved. Earlier versions of this entry (before Oct 08, 2026) said the order's number was not yet published.

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “US Executive Order 14434 replaces 'Artificial Intelligence' with 'Super Intelligence' in executive-branch documents,” The Docket, entry us-fed-eo-super-intelligence-2026-09-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-eo-super-intelligence-2026-09-29.

White House and six AI company leaders sign a frontier-model safety accord

Signed by the President and the heads of Google, Anthropic, Meta, OpenAI, xAI and NVIDIA. Each company commits to internal controls monitoring its models' capabilities and alignment, an internal team overseeing those controls, partnerships with independent external auditors, and an independent committee overseeing the auditors' reports. Voluntary; no penalties reported.

What it asks to be measured. Ongoing monitoring of model capabilities and alignment; reports by internal teams and independent external auditors to an independent committee.

Source: Nextgov/FCW (secondary). Checked Oct 08, 2026.

Limits: No White House or company text of the accord was retrieved; the White House fact sheet on the same day does not mention it. Summary rests on Nextgov's report.

Cite. Protocol STAMP, “White House and six AI company leaders sign a frontier-model safety accord,” The Docket, entry us-fed-frontier-accord-2026-09-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-frontier-accord-2026-09-29.

President Trump at the UN General Assembly rejects a global AI control scheme and renames AI 'Super Intelligence'

In his address to the UN General Assembly, the President said the United States rejects any attempt to construct a globalist scheme of control for AI, and named the technology 'Super Intelligence'. Executive Order 14434 of Sep 29, 2026 applied the name across the executive branch.

What it asks to be measured. None stated. Bears on international AI governance and on the definitions later evaluation rules use.

“totally rejects any attempt to construct a globalist scheme of control”Donald J. Trump, President of the United States, at the UN General Assembly, official White House release

“hereinafter officially called ‘Super Intelligence.’”Donald J. Trump, President of the United States, at the UN General Assembly, official White House release

Source: The White House (release of the address) · The White House (Executive Order 14434). Checked Oct 10, 2026.

Limits: The White House release gives excerpts of the address with video links, not a full transcript; the UN's own record was not retrieved.

Cite. Protocol STAMP, “President Trump at the UN General Assembly rejects a global AI control scheme and renames AI 'Super Intelligence',” The Docket, entry us-fed-trump-unga-2026-09-22, last checked Oct 10, 2026, https://protocolstamp.com/docket/#us-fed-trump-unga-2026-09-22.

California Executive Order N-9-26: accelerating independent oversight

Directs the Government Operations Agency, with the Office of Emergency Services, to speed up implementation of SB 813 and AB 1405 and to convene outside experts within two months. Measures under consideration include placing independent verification organizations onsite at frontier AI developers, verified safety frameworks, an emergency shutoff mechanism for frontier models, and counting loss-of-control events as critical safety incidents.

What it asks to be measured. Regular onsite audits by independent verifiers; verification of developers' safety frameworks and risk assessments (under consideration, not yet required).

Source: Office of the Governor of California · Office of the Governor of California. Checked Oct 08, 2026.

Limits: Summarised from the release; the signed order's PDF was located but not read in full.

Cite. Protocol STAMP, “California Executive Order N-9-26: accelerating independent oversight,” The Docket, entry us-ca-eo-n-9-26-2026-09-18, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-eo-n-9-26-2026-09-18.

California SB 813: independent verification organizations

Signed and chaptered (Chapter 179). Creates a state framework under which the Government Operations Agency designates 'independent verification organizations' (IVOs): AI auditors with demonstrated expertise in assessing the risks of AI systems and models. The agency must set application requirements, qualification criteria and suspension procedures by Jan 01, 2028.

What it asks to be measured. IVOs must identify the benchmarks, metrics and methodologies behind each assessment, stay independent of the party assessed (no payment conditioned on results), and file annual reports to the agency and the Legislature.

Source: California Legislature · Office of the Governor of California. Checked Oct 08, 2026.

Cite. Protocol STAMP, “California SB 813: independent verification organizations,” The Docket, entry us-ca-sb813-2026-09-09, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-sb813-2026-09-09.

California AB 1405: state registry for AI auditors

Signed the same day as SB 813. Creates a state registry for AI auditors with independence and transparency standards.

What it asks to be measured. Who may audit AI systems, and the independence and transparency standards an auditor must meet to be registered.

Source: Office of the Governor of California. Checked Oct 08, 2026.

Limits: Summarised from the Governor's release; the bill text itself was not retrieved for this entry.

Cite. Protocol STAMP, “California AB 1405: state registry for AI auditors,” The Docket, entry us-ca-ab1405-2026-09-09, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-ab1405-2026-09-09.

Anthropic: future Claude models will watermark text, to meet the EU AI Act's marking duty

Anthropic says future Claude models will produce text carrying an imperceptible watermark, applied globally because it cannot yet be scoped by region, and that it will attach C2PA provenance data to files. A detection API is in private preview for eligible organizations such as regulators and researchers. Models launched before Aug 02, 2026 are in a transition period.

What it asks to be measured. Whether a text was written or heavily edited by Claude, checked by a key-holding detector; provenance of files via C2PA.

Source: Anthropic. Checked Oct 08, 2026.

Limits: Anthropic's own account; the method has not been independently evaluated here. It says other major developers signed the same EU Code of Practice; their marks are not covered by this entry.

Cite. Protocol STAMP, “Anthropic: future Claude models will watermark text, to meet the EU AI Act's marking duty,” The Docket, entry eu-anthropic-claude-watermark-2026-08-14, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-anthropic-claude-watermark-2026-08-14.

EU AI Act transparency obligations (Article 50) apply

From Aug 02, 2026, providers of generative AI systems must mark outputs in a machine-readable way, and deployers must disclose deepfakes and certain AI-generated text. The 2026 Omnibus did not move these dates, which also brought most remaining AI Act provisions into application.

What it asks to be measured. Machine-readable marking and detectability of AI-generated content; disclosure duties for deployers.

Source: European Commission · EUR-Lex (Official Journal). Checked Oct 08, 2026.

Limits: Transitional relief for systems already on the market (reported to run to Dec 02, 2026) is from secondary sources.

See explainer: The EU AI Act, from proposal to full application, Labelling AI-generated content: when the duty applies.

Cite. Protocol STAMP, “EU AI Act transparency obligations (Article 50) apply,” The Docket, entry eu-ai-act-art50-applies-2026-08-02, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-act-art50-applies-2026-08-02.

EU Digital Omnibus on AI: high-risk obligations deferred

The AI Omnibus amending the AI Act entered into force on Jul 27, 2026. High-risk obligations for stand-alone (Annex III) systems in areas such as biometrics, employment, education and critical infrastructure now apply from Dec 02, 2027; for AI embedded in regulated products (Annex I), from Aug 02, 2028. General-purpose AI model obligations have applied since Aug 02, 2025.

What it asks to be measured. Unchanged in kind: risk management, testing, accuracy and robustness requirements for high-risk systems now bind later; GPAI evaluation duties already apply.

Source: European Commission · EUR-Lex (Official Journal). Checked Oct 08, 2026.

Limits: Dates from the Commission's page. The regulation number (2026/1744) and Official Journal date (Jul 24, 2026) come from secondary sources; EUR-Lex did not render to our fetcher.

See explainer: The EU AI Act, from proposal to full application.

Cite. Protocol STAMP, “EU Digital Omnibus on AI: high-risk obligations deferred,” The Docket, entry eu-omnibus-ai-2026-07-27, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-omnibus-ai-2026-07-27.

Illinois Artificial Intelligence Safety Measures Act (SB 315, Public Act 104-0538)

Became Public Act 104-0538 on Jul 06, 2026, effective Jan 01, 2027. Large frontier developers must publish and annually update a frontier AI framework (catastrophic-risk assessment, mitigations, governance, cybersecurity, third-party evaluations, internal-use risks), publish transparency reports before deploying new or substantially modified frontier models, undergo annual independent third-party audits, and report critical safety incidents.

What it asks to be measured. Annual independent third-party audits; published frontier AI frameworks; pre-deployment transparency reports; critical safety incident reporting.

Source: Illinois General Assembly. Checked Oct 08, 2026.

Limits: Some duties reportedly start Jan 01, 2028 (secondary sources); the act text was not read in full.

See explainer: US state frontier-AI laws, side by side.

Cite. Protocol STAMP, “Illinois Artificial Intelligence Safety Measures Act (SB 315, Public Act 104-0538),” The Docket, entry us-il-sb315-2026-07-06, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-il-sb315-2026-07-06.

EU Code of Practice on Transparency of AI-generated Content published

Final code published at a closing plenary on Jun 10, 2026, to help providers and deployers meet the AI Act's Article 50 duties to mark and label AI-generated content. Adherence is voluntary; the Article 50 obligations themselves apply from Aug 02, 2026.

What it asks to be measured. Machine-readable marking of AI-generated output by providers; labelling of deepfakes and certain AI-generated text by deployers.

Source: European Commission. Checked Oct 08, 2026.

Cite. Protocol STAMP, “EU Code of Practice on Transparency of AI-generated Content published,” The Docket, entry eu-code-ai-generated-content-2026-06-10, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-code-ai-generated-content-2026-06-10.

NSPM-11: Artificial Intelligence in the National Security Enterprise

A national security presidential memorandum directs the Department of War, the intelligence community and other national security agencies to speed AI adoption, under four pillars: adoption, adaptation, assurance and accountability. It rescinds and replaces the 2024 memorandum NSM-25, bars commercial vendors from disabling, degrading or modifying an AI system warfighters depend on without government approval, and orders DoD Directive 3000.09 on autonomy in weapon systems rewritten within 90 days.

What it asks to be measured. Assurance of AI used for national security; an AI governance policy for national security systems and a rewritten Directive 3000.09 within 90 days; procurement processes updated before Oct 03, 2026 so models from several vendors can be onboarded.

Source: The White House (text as reproduced by GlobalSecurity.org) (secondary) · Small Wars Journal (secondary) · The Next Web (secondary). Checked Oct 09, 2026.

Limits: The memorandum was read only as reproduced by a third party. The applies date is the 90-day deadline as reported. Whether the rewritten Directive 3000.09 was issued was not confirmed. Sources were found by web search on Oct 09, 2026 and could not be opened from our fetcher; the summary rests on the search results and their quoted text, not a full read.

Cite. Protocol STAMP, “NSPM-11: Artificial Intelligence in the National Security Enterprise,” The Docket, entry us-fed-nspm-11-2026-06-05, last checked Oct 09, 2026, https://protocolstamp.com/docket/#us-fed-nspm-11-2026-06-05.

US Executive Order 14409: Promoting Advanced Artificial Intelligence Innovation and Security

Directs cyber-defence upgrades for federal and national-security systems within 30 days, an AI cybersecurity clearinghouse run by CISA in voluntary collaboration with industry and critical-infrastructure operators, and the design of a voluntary framework with AI developers. States that nothing in the section authorises a mandatory licensing, preclearance or permitting requirement for AI models. Issued alongside NSPM-11.

What it asks to be measured. A voluntary framework for government access to advanced models (per law-firm summaries, up to 30 days pre-release for models with advanced cyber capabilities identified by classified benchmarking).

Source: The White House. Checked Oct 08, 2026.

Limits: The 30-day pre-release access and classified benchmarking details are from law-firm summaries; the order's definitions section was not read in full.

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “US Executive Order 14409: Promoting Advanced Artificial Intelligence Innovation and Security,” The Docket, entry us-fed-eo14409-2026-06-02, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-eo14409-2026-06-02.

Connecticut SB 5 (Public Act 26-15), 'An Act Concerning Online Safety', with AI provisions

Signed by the Governor on May 27, 2026. An omnibus act whose AI provisions, per law-firm summaries, cover AI in employment decisions, AI companions and consumer generative-AI tools, frontier-model developers, and platforms used by minors, with staggered effective dates from Oct 01, 2026 to Jan 01, 2028. Commentators have called it the 'Connecticut Artificial Intelligence Responsibility and Transparency Act'; the Legislature's title is 'An Act Concerning Online Safety'.

What it asks to be measured. Notices for AI-driven employment decisions; duties for AI companion operators and frontier developers (details from secondary summaries).

Source: Connecticut General Assembly. Checked Oct 08, 2026.

Limits: Provision-level detail and effective dates are from law-firm summaries; the act's text was not read.

Cite. Protocol STAMP, “Connecticut SB 5 (Public Act 26-15), 'An Act Concerning Online Safety', with AI provisions,” The Docket, entry us-ct-sb5-2026-05-27, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ct-sb5-2026-05-27.

Colorado SB26-189 repeals and replaces the Colorado AI Act

Signed May 14, 2026. Repeals and re-enacts the provisions of SB 24-205, the 2024 Colorado AI Act, around 'automated decision-making technology' used in consequential decisions such as employment, housing and financial services, in place of 'high-risk artificial intelligence systems'.

What it asks to be measured. Disclosure and consumer rights around automated decisions; the impact-assessment model of the 2024 Act is replaced.

Source: Colorado General Assembly. Checked Oct 08, 2026.

Limits: The Jan 01, 2027 date for the new obligations comes from secondary sources; the bill page lists the act's effective date as May 14, 2026.

Cite. Protocol STAMP, “Colorado SB26-189 repeals and replaces the Colorado AI Act,” The Docket, entry us-co-sb26-189-2026-05-14, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-co-sb26-189-2026-05-14.

UK King's Speech 2026: no standalone AI bill

The King's Speech did not announce a dedicated AI bill. It announced a Regulating for Growth Bill with cross-economy sandboxing powers that would let businesses test products, including AI, in controlled environments. Sector regulators keep their AI remits.

What it asks to be measured. None stated for AI evaluation; sandbox testing powers proposed.

Source: House of Commons Library · IAPP (secondary) · Lewis Silkin LLP (law firm) (secondary). Checked Oct 08, 2026.

Limits: The Commons Library briefing refused our fetcher (403). The IAPP report confirms no AI bill; the Regulating for Growth Bill and its AI sandbox powers are from Lewis Silkin's note, which quotes the government's briefing notes.

Cite. Protocol STAMP, “UK King's Speech 2026: no standalone AI bill,” The Docket, entry uk-kings-speech-2026-05-13, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-kings-speech-2026-05-13.

New York RAISE Act chapter amendment

A chapter amendment signed Mar 27, 2026 is reported to be the final form of the RAISE Act: it aligns key definitions with California's Transparency in Frontier AI Act while keeping a 72-hour incident-reporting window (California's is 15 days), with obligations from Jan 01, 2027.

What it asks to be measured. As above; definitions aligned with California's frontier-model law.

Source: Wiley Rein LLP (law firm) (secondary) · Cooley LLP (law firm) (secondary). Checked Oct 08, 2026.

Limits: No primary source retrieved for the amendment; two law-firm analyses agree on the date and terms.

Cite. Protocol STAMP, “New York RAISE Act chapter amendment,” The Docket, entry us-ny-raise-amend-2026-03-27, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ny-raise-amend-2026-03-27.

Ukraine publishes a draft AI Development Strategy to 2030

The Ministry of Digital Transformation published a draft AI Development Strategy to 2030 and an operational plan for 2026-2028, after a two-stage regulatory roadmap, a white paper, a voluntary code of conduct (December 2024) and a series of sector recommendations, as Ukraine prepares legislation aligned with the EU AI Act.

What it asks to be measured. Voluntary code and sector recommendations now; legislation aligned with the EU AI Act planned.

Source: Ministry of Digital Transformation of Ukraine (Centre for Responsible AI). Checked Oct 09, 2026.

Limits: The issuer dates this document to the month only; the entry is placed on the first day of that month. The EU-alignment aim is from the Verkhovna Rada research service and press coverage.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Ukraine publishes a draft AI Development Strategy to 2030,” The Docket, entry ua-ai-strategy-draft-2026-03-01, last checked Oct 09, 2026, https://protocolstamp.com/docket/#ua-ai-strategy-draft-2026-03-01.

New Delhi Declaration on AI Impact (AI Impact Summit, India)

Adopted at the close of the AI Impact Summit held in New Delhi on Feb 18, 2026 and Feb 19, 2026, per the Government of India. Secondary reports describe seven pillars ('chakras'), one on secure and trusted AI, and endorsement counts between 86 and 92 countries and organisations as more joined.

What it asks to be measured. None specific; a pillar on secure and trusted AI systems.

Source: National e-Governance Division, Government of India. Checked Oct 08, 2026.

Limits: The issuer page confirms dates and adoption only; pillars and signatory counts are from secondary reports and vary by date.

Cite. Protocol STAMP, “New Delhi Declaration on AI Impact (AI Impact Summit, India),” The Docket, entry intl-new-delhi-declaration-2026-02-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#intl-new-delhi-declaration-2026-02-19.

India IT Rules amended for synthetically generated information

MeitY notified the IT (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 by Gazette notification G.S.R. 120(E) dated Feb 10, 2026, in force Feb 20, 2026, after a draft in October 2025. Defines 'synthetically generated information' in Rule 2(1)(wa).

What it asks to be measured. Visible labels and embedded metadata on synthetic content; user declarations and technical verification by significant social media intermediaries.

Source: Vikaspedia (Government of India portal run by C-DAC for MeitY) (secondary) · Data Security Council of India (secondary). Checked Oct 08, 2026.

Limits: The Gazette PDF itself was not retrieved; details from a government information portal and DSCI. A draft Second Amendment followed in 2026.

See explainer: Labelling AI-generated content: when the duty applies, Where countries stand.

Cite. Protocol STAMP, “India IT Rules amended for synthetically generated information,” The Docket, entry in-it-rules-sgi-2026-02-10, last checked Oct 08, 2026, https://protocolstamp.com/docket/#in-it-rules-sgi-2026-02-10.

REAIM 2026 in A Coruña adopts 'Pathways to Action' on military AI

The third summit on Responsible AI in the Military Domain, held by Spain's foreign and defence ministries over two days ending Feb 05, 2026, adopted a declaration on putting responsible development, deployment and use of military AI into practice, including human responsibility over AI-enabled weapons and command chains and sharing information on national oversight. The United States and China did not endorse it. The document stays open to further states.

What it asks to be measured. States share information on their national oversight of military AI; human responsibility and accountability over AI-enabled weapons.

Source: Ministerio de Asuntos Exteriores, Spain · Ministerio de Asuntos Exteriores, Spain (press release) · Reuters, via U.S. News & World Report (secondary). Checked Oct 10, 2026.

Limits: Endorsement counts differ: 34 or 35 of about 85 attending states at the summit, as reported; more than 40 per the organisers later, and about 45 on the ministry's list as read in a search result. Sources were found by web search on Oct 09, 2026 and could not be opened from our fetcher; the summary rests on the search results and their quoted text, not a full read.

Cite. Protocol STAMP, “REAIM 2026 in A Coruña adopts 'Pathways to Action' on military AI,” The Docket, entry intl-reaim-coruna-2026-02-05, last checked Oct 10, 2026, https://protocolstamp.com/docket/#intl-reaim-coruna-2026-02-05.

South Korea AI Basic Act and Enforcement Decree take effect

The AI Basic Act and its Enforcement Decree took effect on Jan 22, 2026, combining support for the AI industry with baseline duties of trust, safety, transparency and user protection. The ministry (MSIT) is deferring fact-finding investigations and fines for at least a year except in exceptional cases.

What it asks to be measured. Advance notice to users of high-impact and generative AI; labelling of synthetic content; risk management plans, explainability and human-oversight documentation for high-impact systems.

Source: US International Trade Administration (secondary) · US Library of Congress, Global Legal Monitor (secondary). Checked Oct 08, 2026.

Limits: Official US-government summaries, not the Korean issuer's text; the Library of Congress page refused our fetcher (403).

See explainer: Labelling AI-generated content: when the duty applies, Where countries stand.

Cite. Protocol STAMP, “South Korea AI Basic Act and Enforcement Decree take effect,” The Docket, entry kr-ai-basic-act-2026-01-22, last checked Oct 08, 2026, https://protocolstamp.com/docket/#kr-ai-basic-act-2026-01-22.

US Defense Secretary memo: an 'AI-first' department and 'any lawful use' terms in AI contracts

A memo from the Secretary of Defense, styled Secretary of War, directs the department to adopt AI across all components and to use models without vendor usage-policy limits. The Under Secretary for Acquisition and Sustainment is to add standard 'any lawful use' language to every contract that procures AI services within 180 days. The policy led to the department's dispute with Anthropic.

What it asks to be measured. Benchmarks for model objectivity from the Chief Digital and AI Office within 90 days, as reported; expanded AI compute at the tactical edge. No public test or pass mark.

Source: DefenseScoop (secondary) · Holland & Knight (secondary) · Lawfare (secondary). Checked Oct 09, 2026.

Limits: No department text of the memo was retrieved. The applies date is 180 days after Jan 09, 2026, computed from the reported deadline. Sources were found by web search on Oct 09, 2026 and could not be opened from our fetcher; the summary rests on the search results and their quoted text, not a full read.

Cite. Protocol STAMP, “US Defense Secretary memo: an 'AI-first' department and 'any lawful use' terms in AI contracts,” The Docket, entry us-fed-dow-ai-first-memo-2026-01-09, last checked Oct 09, 2026, https://protocolstamp.com/docket/#us-fed-dow-ai-first-memo-2026-01-09.

Taiwan Artificial Intelligence Fundamental Act (AI Basic Act)

Passed third reading in the Legislative Yuan on Dec 23, 2025 (the Ministry of Digital Affairs' English name is 'Artificial Intelligence Fundamental Act'); reported promulgated Jan 14, 2026. Sets principles, a national AI strategy committee, and assigns the Ministry of Digital Affairs to develop risk classification frameworks.

What it asks to be measured. Risk classification framework; verification and labelling mechanisms for AI.

Source: Ministry of Digital Affairs (Taiwan). Checked Oct 08, 2026.

Limits: Promulgation date from secondary sources.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Taiwan Artificial Intelligence Fundamental Act (AI Basic Act),” The Docket, entry tw-ai-basic-act-2025-12-23, last checked Oct 08, 2026, https://protocolstamp.com/docket/#tw-ai-basic-act-2025-12-23.

New York RAISE Act signed (S6953B/A6453B)

Requires frontier AI developers to create and publish safety protocol information and to report critical-harm incidents to the state within 72 hours. Oversight sits with a new office in the Department of Financial Services, which assesses large frontier developers and issues annual transparency reports. Penalties up to $1 million for a first violation and $3 million for later ones.

What it asks to be measured. Published safety protocols; incident reports within 72 hours; state assessment of large frontier developers.

“setting a strong and sensible standard for frontier AI safety”Kathy Hochul, Governor of New York, official press release

Source: Office of the Governor of New York. Checked Oct 08, 2026.

Limits: The Jan 01, 2027 date is the amended law's, from the entry below.

See explainer: US state frontier-AI laws, side by side.

Cite. Protocol STAMP, “New York RAISE Act signed (S6953B/A6453B),” The Docket, entry us-ny-raise-2025-12-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ny-raise-2025-12-19.

US Executive Order 14365: national policy framework and state AI laws

Directs the Attorney General to form an AI Litigation Task Force to challenge state AI laws that conflict with federal policy; the Commerce Secretary to identify 'onerous' state laws within 90 days; agencies to consider conditioning BEAD and discretionary grants; the FCC to begin a proceeding on a federal AI reporting standard; and officials to draft preemptive federal legislation, with carve-outs including child safety.

What it asks to be measured. A proposed federal AI reporting standard (FCC proceeding); otherwise none stated. Bears on whether state evaluation and audit laws, such as California's, remain enforceable.

Source: The White House · US Government Publishing Office (Federal Register). Checked Oct 08, 2026.

Limits: An executive order cannot itself overturn state law; that requires Congress or the courts.

See explainer: US federal AI policy, 2020 to 2026, Where countries stand.

Cite. Protocol STAMP, “US Executive Order 14365: national policy framework and state AI laws,” The Docket, entry us-fed-eo14365-2025-12-11, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-eo14365-2025-12-11.

Vietnam Law on Artificial Intelligence (No. 134/2025/QH15)

Passed by the National Assembly on Dec 10, 2025 (429 of 434 votes); effective Mar 01, 2026. Risk-based, three tiers; transition to Mar 01, 2027 (Sep 01, 2027 for health, education and finance).

What it asks to be measured. Risk classification and conformity duties for higher-risk AI; labelling of AI-generated content.

Source: VnEconomy (secondary) · Baker McKenzie (law firm) (secondary). Checked Oct 08, 2026.

Limits: No Vietnamese government text retrieved.

See explainer: Labelling AI-generated content: when the duty applies, Where countries stand.

Cite. Protocol STAMP, “Vietnam Law on Artificial Intelligence (No. 134/2025/QH15),” The Docket, entry vn-ai-law-2025-12-10, last checked Oct 08, 2026, https://protocolstamp.com/docket/#vn-ai-law-2025-12-10.

Australia National AI Plan: mandatory guardrails paused, AI Safety Institute funded

Published Dec 02, 2025. Says regulation will build on Australia's existing legal and regulatory frameworks rather than a standalone AI act, and establishes an AI Safety Institute to monitor, test and share information on emerging AI capabilities, risks and harms.

What it asks to be measured. AI Safety Institute to monitor and test emerging AI risks; no binding evaluation duties.

Source: Department of Industry, Science and Resources (Australia) · Department of Industry, Science and Resources (Australia). Checked Oct 08, 2026.

Limits: The A$29.9 million AISI figure and the description of mandatory guardrails as paused come from press coverage, not the plan pages read.

See explainer: Government AI testing bodies, Where countries stand.

Cite. Protocol STAMP, “Australia National AI Plan: mandatory guardrails paused, AI Safety Institute funded,” The Docket, entry au-national-ai-plan-2025-12-02, last checked Oct 08, 2026, https://protocolstamp.com/docket/#au-national-ai-plan-2025-12-02.

Digital Omnibus on AI proposed

Commission proposal of Nov 19, 2025 for 'targeted simplification measures' to the AI Act's implementation. As adopted in 2026, it moved the high-risk dates to Dec 02, 2027 (Annex III) and Aug 02, 2028 (Annex I); see the July 2026 entry.

What it asks to be measured. Timing of high-risk testing and conformity duties.

Source: European Commission. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Digital Omnibus on AI proposed,” The Docket, entry eu-digital-omnibus-proposal-2025-11-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-digital-omnibus-proposal-2025-11-19.

Kazakhstan Law on Artificial Intelligence signed

Signed by President Tokayev on Nov 17, 2025; the first comprehensive AI law in Central Asia. Risk tiers, prohibited manipulative uses, and mandatory labelling of AI-generated content.

What it asks to be measured. Risk classification with special control of high-risk systems; labelling of generated text, images, audio and video.

Source: The Astana Times (secondary) · Kazakhstan Today (secondary). Checked Oct 08, 2026.

Limits: No Akorda (presidential) or Adilet text retrieved.

See explainer: Labelling AI-generated content: when the duty applies, Where countries stand.

Cite. Protocol STAMP, “Kazakhstan Law on Artificial Intelligence signed,” The Docket, entry kz-ai-law-2025-11-17, last checked Oct 08, 2026, https://protocolstamp.com/docket/#kz-ai-law-2025-11-17.

India AI Governance Guidelines

Released by MeitY on Nov 05, 2025 (per DSCI). A governance framework that relies on existing laws rather than a new AI statute, per officials quoted in coverage.

What it asks to be measured. Recommends techno-legal tools, incident databases and voluntary frameworks; no binding evaluation duties.

Source: Data Security Council of India (secondary). Checked Oct 08, 2026.

Limits: Primary MeitY/PIB document not retrieved; date differs by one day across sources (Nov 04, 2025 or Nov 05, 2025).

Cite. Protocol STAMP, “India AI Governance Guidelines,” The Docket, entry in-ai-governance-guidelines-2025-11-05, last checked Oct 08, 2026, https://protocolstamp.com/docket/#in-ai-governance-guidelines-2025-11-05.

China amends the Cybersecurity Law to add AI provisions

Adopted by the NPC Standing Committee on Oct 28, 2025; effective Jan 01, 2026. A new article supports AI research and infrastructure and calls for AI ethics rules, risk monitoring and safety evaluation.

What it asks to be measured. AI risk monitoring and safety evaluation (framework level; details left to implementing rules).

Source: Cyberspace Administration of China · Xinhua (state news agency) (secondary) · China Daily (NPC subsite) (secondary). Checked Oct 08, 2026.

Limits: The NPC's own page was not reached; the CAC source is an expert commentary on the amendment published by the regulator.

Cite. Protocol STAMP, “China amends the Cybersecurity Law to add AI provisions,” The Docket, entry cn-cybersecurity-law-amend-2025-10-28, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-cybersecurity-law-amend-2025-10-28.

California SB 243: companion chatbots

Signed Oct 13, 2025, effective Jan 01, 2026. Operators of companion chatbots must disclose that the chatbot is AI, maintain protocols for suicidal ideation and self-harm, add protections for minors, and report annually from July 2027.

What it asks to be measured. Annual reports to the Office of Suicide Prevention on crisis-referral protocols and counts.

Source: California Legislature. Checked Oct 08, 2026.

Cite. Protocol STAMP, “California SB 243: companion chatbots,” The Docket, entry us-ca-sb243-2025-10-13, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-sb243-2025-10-13.

Italy Law 132/2025 on artificial intelligence

Published in the Gazzetta Ufficiale on Sep 25, 2025 and in force Oct 10, 2025; the first national AI law in the EU, complementing the AI Act. Designates AgID and ACN as national AI authorities and adds a deepfake offence.

What it asks to be measured. Sectoral transparency and human-oversight rules (health, employment, professions, public administration).

Source: Gazzetta Ufficiale della Repubblica Italiana. Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Italy Law 132/2025 on artificial intelligence,” The Docket, entry it-law-132-2025-10-10, last checked Oct 08, 2026, https://protocolstamp.com/docket/#it-law-132-2025-10-10.

California Civil Rights Council regulations on automated-decision systems in employment

Effective Oct 01, 2025. Clarifies that use of automated-decision systems in hiring and employment can violate the Fair Employment and Housing Act and that anti-bias testing is relevant to defences; four-year record retention.

What it asks to be measured. Anti-bias testing of automated-decision systems (relevant evidence); record retention of ADS data.

Source: California Civil Rights Department. Checked Oct 08, 2026.

Cite. Protocol STAMP, “California Civil Rights Council regulations on automated-decision systems in employment,” The Docket, entry us-ca-crd-ads-regs-2025-10-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-crd-ads-regs-2025-10-01.

California SB 53: Transparency in Frontier Artificial Intelligence Act

Signed Sep 29, 2025, effective Jan 01, 2026. Large frontier developers must publish a frontier AI framework and transparency reports, and report critical safety incidents to the Office of Emergency Services within 15 days (24 hours if there is imminent risk of death or serious injury). 'Frontier model' means training compute above 10^26 operations; 'large' means over $500,000,000 in annual revenue.

What it asks to be measured. Published frontier AI framework covering catastrophic-risk assessment and third-party evaluation use; critical safety incident reports; summaries of catastrophic-risk assessments.

Source: California Legislature · Office of the Governor of California. Checked Oct 08, 2026.

See explainer: US state frontier-AI laws, side by side, Training-compute thresholds that trigger AI rules, Where countries stand.

Cite. Protocol STAMP, “California SB 53: Transparency in Frontier Artificial Intelligence Act,” The Docket, entry us-ca-sb53-2025-09-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-sb53-2025-09-29.

California CPPA regulations on automated decision-making technology, risk assessments and cybersecurity audits

Approved by the Office of Administrative Law in September 2025, effective Jan 01, 2026 with phased compliance: ADMT duties from Jan 01, 2027; risk-assessment attestations from 2028; cybersecurity audits phased from 2028.

What it asks to be measured. Pre-use notices and opt-out for significant decisions made by ADMT; risk assessments submitted to the CPPA; independent cybersecurity audits.

Source: California Privacy Protection Agency. Checked Oct 08, 2026.

Cite. Protocol STAMP, “California CPPA regulations on automated decision-making technology, risk assessments and cybersecurity audits,” The Docket, entry us-ca-cppa-admt-2025-09-23, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-cppa-admt-2025-09-23.

UN General Assembly resolution 79/325: Scientific Panel on AI and Global Dialogue on AI Governance

Adopted by consensus. Sets terms of reference for an Independent International Scientific Panel on AI (40 members) and a Global Dialogue on AI Governance, both agreed in principle in the 2024 Global Digital Compact.

What it asks to be measured. A standing scientific panel to assess AI opportunities, risks and impacts and report annually.

Source: United Nations (Office of the President of the General Assembly). Checked Oct 08, 2026.

Limits: Panel report cadence taken from secondary summaries of draft text.

Cite. Protocol STAMP, “UN General Assembly resolution 79/325: Scientific Panel on AI and Global Dialogue on AI Governance,” The Docket, entry un-ga-res-79-325-2025-08-26, last checked Oct 08, 2026, https://protocolstamp.com/docket/#un-ga-res-79-325-2025-08-26.

Chile's Chamber of Deputies approves the AI regulation bill in general

The Chamber approved in general bills 15869 and 16821, which regulate the development, implementation and use of AI, by 63 votes to 47 with 6 abstentions; provisions needing a special quorum were rejected and the bill returned to committee. It assigns supervision to the Personal Data Protection Agency and uses a risk-based approach.

What it asks to be measured. Risk classification of AI systems; supervision by the data-protection agency (proposed).

Source: Cámara de Diputadas y Diputados de Chile. Checked Oct 09, 2026.

Limits: Status after August 2025 (committee report, Senate) was not checked.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Chile's Chamber of Deputies approves the AI regulation bill in general,” The Docket, entry cl-ai-bill-2025-08-04, last checked Oct 09, 2026, https://protocolstamp.com/docket/#cl-ai-bill-2025-08-04.

America's AI Action Plan and executive orders of Jul 23, 2025

White House plan with over 90 federal actions. Same day: EO 14319 'Preventing Woke AI in the Federal Government' requiring federally procured LLMs to meet 'unbiased AI principles', plus orders on data-centre permitting and AI exports.

What it asks to be measured. CAISI evaluations of frontier models for national-security risks; an AI evaluations ecosystem; procurement conditions on LLM neutrality and truthfulness.

Source: The White House · The White House. Checked Oct 08, 2026.

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “America's AI Action Plan and executive orders of Jul 23, 2025,” The Docket, entry us-fed-ai-action-plan-2025-07-23, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-ai-action-plan-2025-07-23.

Commission guidelines on obligations for general-purpose AI model providers and training-data summary template

Guidelines (Jul 18, 2025) clarify who is a GPAI provider, compute thresholds, and systemic-risk classification. A mandatory public template for summarising training content followed on Jul 24, 2025.

What it asks to be measured. Training compute (10^23 FLOP indicative GPAI threshold; 10^25 FLOP presumption of systemic risk); public summary of training data.

Source: European Commission · European Commission. Checked Oct 08, 2026.

See explainer: Training-compute thresholds that trigger AI rules.

Cite. Protocol STAMP, “Commission guidelines on obligations for general-purpose AI model providers and training-data summary template,” The Docket, entry eu-gpai-guidelines-2025-07-18, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-gpai-guidelines-2025-07-18.

General-Purpose AI Code of Practice published

Three chapters (transparency, copyright, safety and security) drafted by independent experts. Signing is voluntary and offers a way to show compliance with the AI Act's GPAI obligations, which applied from Aug 02, 2025.

What it asks to be measured. Model documentation form; for systemic-risk models: state-of-the-art model evaluations, systemic-risk assessment, external evaluation in some cases, and serious-incident reporting.

Source: European Commission. Checked Oct 08, 2026.

See explainer: The EU AI Act, from proposal to full application.

Cite. Protocol STAMP, “General-Purpose AI Code of Practice published,” The Docket, entry eu-gpai-code-2025-07-10, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-gpai-code-2025-07-10.

US Senate votes 99-1 to strip a state AI-law moratorium from the budget bill

During debate on the reconciliation bill later enacted as the One Big Beautiful Bill Act, the Senate adopted Blackburn Amendment No. 2814, 'to strike the section relating to support for artificial intelligence', by 99 votes to 1. The stricken section would have conditioned federal funds on states not enforcing AI laws.

What it asks to be measured. Bears on whether state AI evaluation and audit laws remain enforceable.

Source: United States Senate (roll call vote). Checked Oct 08, 2026.

Limits: What the stricken section did is from contemporaneous reporting; the roll call gives only the amendment's stated purpose.

Cite. Protocol STAMP, “US Senate votes 99-1 to strip a state AI-law moratorium from the budget bill,” The Docket, entry us-fed-senate-moratorium-vote-2025-07-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-senate-moratorium-vote-2025-07-01.

Texas Responsible Artificial Intelligence Governance Act (HB 149)

Signed Jun 22, 2025, effective Jan 01, 2026. Prohibits certain intentional harmful AI uses, requires government agencies to disclose AI interactions, creates a regulatory sandbox and an AI council; enforcement by the Attorney General.

What it asks to be measured. Disclosure of AI interaction by government agencies and health-care providers; safe-harbour for compliance with the NIST AI RMF.

Source: Texas Legislature Online. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Texas Responsible Artificial Intelligence Governance Act (HB 149),” The Docket, entry us-tx-traiga-2025-06-22, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-tx-traiga-2025-06-22.

UK Data (Use and Access) Act 2025: automated decision-making rules reformed

Royal Assent Jun 19, 2025. Replaces UK GDPR Article 22 with new rules permitting solely automated decisions with significant effects in more cases, subject to safeguards; stricter limits remain for special-category data. Commenced in stages.

What it asks to be measured. Safeguards for automated decisions: information to the individual, ability to contest and obtain human intervention.

Source: legislation.gov.uk · legislation.gov.uk. Checked Oct 08, 2026.

Limits: Commencement dates for the ADM provisions are set by regulations; check before relying on them.

Cite. Protocol STAMP, “UK Data (Use and Access) Act 2025: automated decision-making rules reformed,” The Docket, entry uk-data-use-access-act-2025-06-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-data-use-access-act-2025-06-19.

California Report on Frontier AI Policy

Final report of the Joint California Policy Working Group on AI Frontier Models (co-led by Fei-Fei Li, Mariano-Florentino Cuéllar and Jennifer Tour Chayes), convened after the SB 1047 veto.

What it asks to be measured. Recommends third-party risk assessment, transparency, whistleblower protections and adverse-event reporting.

Source: Office of the Governor of California · Office of the Governor of California (PDF). Checked Oct 08, 2026.

Limits: A policy report, not a law. The working group's own domain (cafrontieraigov.org) has lapsed and served unrelated gambling content on Oct 08, 2026; do not cite it.

Cite. Protocol STAMP, “California Report on Frontier AI Policy,” The Docket, entry us-ca-frontier-report-2025-06-17, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-frontier-report-2025-06-17.

Japan Act on Promotion of Research, Development and Utilization of AI-Related Technologies

Passed by the Diet on May 28, 2025, promulgated Jun 04, 2025 (Act No. 53 of 2025); the AI Strategy Headquarters provisions took effect Sep 01, 2025. A framework law without penalties; the government may investigate and publicise inappropriate AI use.

What it asks to be measured. Government research into AI-related incidents and rights infringements; a national AI Basic Plan.

Source: Japanese Law Translation (Ministry of Justice) · House of Representatives of Japan. Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Japan Act on Promotion of Research, Development and Utilization of AI-Related Technologies,” The Docket, entry jp-ai-promotion-act-2025-06-04, last checked Oct 08, 2026, https://protocolstamp.com/docket/#jp-ai-promotion-act-2025-06-04.

US AI Safety Institute reformed as the Center for AI Standards and Innovation (CAISI)

The Commerce Secretary announced the reform of the AI Safety Institute into CAISI, to work with industry on voluntary agreements and evaluate frontier models for national-security risks.

What it asks to be measured. Voluntary evaluations of frontier models for demonstrable risks such as cybersecurity, biosecurity and chemical weapons; evaluation of adversary AI systems.

Source: US Department of Commerce · US National Institute of Standards and Technology. Checked Oct 08, 2026.

Limits: Commerce.gov served a bot-check page to our fetcher. As of Oct 08, 2026 the NIST page is titled 'Center for Advancing Innovation and Standards for Super Intelligence (CAISSI)'; the date and instrument of that rename were not retrieved.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “US AI Safety Institute reformed as the Center for AI Standards and Innovation (CAISI),” The Docket, entry us-fed-caisi-2025-06-03, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-caisi-2025-06-03.

TAKE IT DOWN Act signed (Public Law 119-12)

Criminalises publishing non-consensual intimate imagery, including AI-generated 'digital forgeries', and requires covered platforms to remove such content within 48 hours of a valid request; platform duties applied one year after enactment.

What it asks to be measured. Notice-and-removal process with a 48-hour removal window.

Source: US Congress (Congress.gov). Checked Oct 08, 2026.

Cite. Protocol STAMP, “TAKE IT DOWN Act signed (Public Law 119-12),” The Docket, entry us-fed-take-it-down-2025-05-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-take-it-down-2025-05-19.

OMB Memoranda M-25-21 and M-25-22 on federal AI use and procurement

Replaced M-24-10 and M-24-18. Keeps Chief AI Officers and minimum risk practices, now for 'high-impact AI', and sets procurement guidance.

What it asks to be measured. Pre-deployment testing, AI impact assessments and ongoing monitoring for high-impact AI; human oversight.

Source: Office of Management and Budget · Office of Management and Budget. Checked Oct 08, 2026.

Cite. Protocol STAMP, “OMB Memoranda M-25-21 and M-25-22 on federal AI use and procurement,” The Docket, entry us-fed-omb-m25-21-2025-04-03, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-omb-m25-21-2025-04-03.

Kenya National Artificial Intelligence Strategy 2025-2030 launched

Launched at the Kenyatta International Convention Centre, Nairobi, on Mar 27, 2025 by the Ministry of Information, Communications and the Digital Economy. Three pillars: AI digital infrastructure, data, and AI research and innovation, with governance and ethics among its enablers. A draft AI and Emerging Technologies Policy followed in 2026.

What it asks to be measured. Governance and ethics frameworks (planned); no evaluation duties.

Source: Ministry of ICT and the Digital Economy (Kenya) · Ministry of ICT and the Digital Economy (Kenya), draft 2026 policy. Checked Oct 09, 2026.

Limits: Pillars are from secondary summaries of the strategy.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Kenya National Artificial Intelligence Strategy 2025-2030 launched,” The Docket, entry ke-ai-strategy-2025-03-27, last checked Oct 09, 2026, https://protocolstamp.com/docket/#ke-ai-strategy-2025-03-27.

China: Measures for Labeling AI-Generated Synthetic Content

Issued by the CAC and three other agencies; effective Sep 01, 2025, with the mandatory national standard GB 45438-2025 on the same date.

What it asks to be measured. Explicit (visible) labels and implicit labels (metadata) on AI-generated text, images, audio, video and virtual scenes; platform detection duties.

Source: Cyberspace Administration of China. Checked Oct 08, 2026.

See explainer: Labelling AI-generated content: when the duty applies.

Cite. Protocol STAMP, “China: Measures for Labeling AI-Generated Synthetic Content,” The Docket, entry cn-labeling-measures-2025-03-14, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-labeling-measures-2025-03-14.

UK AI Safety Institute renamed AI Security Institute

The institute was refocused on serious AI risks with security implications, such as chemical and biological weapons uplift, cyber attacks and crime.

What it asks to be measured. Evaluation focus narrowed to security-relevant capabilities.

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “UK AI Safety Institute renamed AI Security Institute,” The Docket, entry uk-aisi-rename-2025-02-14, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-aisi-rename-2025-02-14.

Colombia approves its National Artificial Intelligence Policy (CONPES 4144)

Approved by the National Council for Economic and Social Policy on Feb 14, 2025, succeeding CONPES 3975 of 2019. Six strategic axes, from ethics and governance to data and infrastructure, risk mitigation and adoption, with more than 100 actions to 2030 led by MinTIC and MinCiencias.

What it asks to be measured. Ethics and governance axis; risk mitigation measures for AI use.

Source: Departamento Nacional de Planeación (Colombia). Checked Oct 09, 2026.

Limits: Investment figure and action count are from DNP summaries, not checked line by line.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Colombia approves its National Artificial Intelligence Policy (CONPES 4144),” The Docket, entry co-conpes-4144-2025-02-14, last checked Oct 09, 2026, https://protocolstamp.com/docket/#co-conpes-4144-2025-02-14.

Switzerland decides to ratify the Council of Europe AI Convention and adapt existing law

The Federal Council chose to ratify the CoE Framework Convention and to make sector-specific amendments to Swiss law rather than adopt a horizontal AI act, with a consultation draft due by end-2026.

What it asks to be measured. Implementing the Convention's risk and impact assessment duties through sectoral law (planned).

Source: Swiss Federal Council. Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Switzerland decides to ratify the Council of Europe AI Convention and adapt existing law,” The Docket, entry ch-coe-ratification-plan-2025-02-12, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ch-coe-ratification-plan-2025-02-12.

Paris AI Action Summit: Statement on Inclusive and Sustainable AI

Issued at the AI Action Summit in Paris; signed by about 60 countries including China and India. The United States and United Kingdom did not sign.

What it asks to be measured. None stated for evaluation; references openness, inclusivity and trustworthy AI.

Source: Élysée (Presidency of France). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Paris AI Action Summit: Statement on Inclusive and Sustainable AI,” The Docket, entry intl-paris-statement-2025-02-11, last checked Oct 08, 2026, https://protocolstamp.com/docket/#intl-paris-statement-2025-02-11.

AI Act prohibitions and AI literacy apply; Commission guidelines on prohibited practices and AI-system definition

Article 5 prohibitions (e.g. social scoring, untargeted facial-image scraping, certain emotion recognition) and the Article 4 AI-literacy duty applied from Feb 02, 2025. The Commission published guidelines on prohibited practices (Feb 04, 2025) and on the definition of an AI system (Feb 06, 2025).

What it asks to be measured. Whether a system meets the AI-system definition; whether a use falls within a prohibited practice.

Source: European Commission · European Commission. Checked Oct 08, 2026.

See explainer: The EU AI Act, from proposal to full application.

Cite. Protocol STAMP, “AI Act prohibitions and AI literacy apply; Commission guidelines on prohibited practices and AI-system definition,” The Docket, entry eu-prohibited-practices-2025-02-02, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-prohibited-practices-2025-02-02.

UK AI Cyber Security Code of Practice

Voluntary code from DSIT setting 13 baseline cyber-security principles across the AI lifecycle; intended as the basis of an ETSI global standard.

What it asks to be measured. Security testing of AI models and systems before release; documentation of data, models and prompts; monitoring.

Source: UK Government (GOV.UK) · UK Government (GOV.UK). Checked Oct 08, 2026.

Cite. Protocol STAMP, “UK AI Cyber Security Code of Practice,” The Docket, entry uk-ai-cyber-code-2025-01-31, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-ai-cyber-code-2025-01-31.

International AI Safety Report 2025

Chaired by Yoshua Bengio and mandated at the Bletchley summit, written by 96 experts nominated by 30 countries, the OECD, EU and UN. Summarises the science on general-purpose AI capabilities, risks and mitigations.

What it asks to be measured. States of evidence on evaluation methods, their limits, and risk-management techniques for general-purpose AI.

Source: UK Government (GOV.UK) · International AI Safety Report (secondary). Checked Oct 08, 2026.

Limits: A scientific report, not a binding instrument.

Cite. Protocol STAMP, “International AI Safety Report 2025,” The Docket, entry intl-ai-safety-report-2025-01-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#intl-ai-safety-report-2025-01-29.

US Executive Order 14179: removing barriers to American leadership in AI

Signed Jan 23, 2025 after EO 14110 was revoked on Jan 20, 2025. Directed a review of actions taken under EO 14110 and an AI Action Plan within 180 days.

What it asks to be measured. Rescinds or revises federal evaluation and reporting requirements adopted under EO 14110.

Source: Federal Register. Checked Oct 08, 2026.

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “US Executive Order 14179: removing barriers to American leadership in AI,” The Docket, entry us-fed-eo14179-2025-01-23, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-eo14179-2025-01-23.

South Korea Framework Act on AI promulgated

Passed by the National Assembly on Dec 26, 2024 and promulgated Jan 21, 2025; took effect Jan 22, 2026 (see existing entry). Second comprehensive AI law after the EU's.

What it asks to be measured. Risk management and explainability for high-impact AI; generative-AI notice and labelling; safety duties for models above a compute threshold.

Source: Korea Law Information Center (Ministry of Government Legislation) · Ministry of Science and ICT. Checked Oct 08, 2026.

Limits: Links are to the official law database search and ministry site, not a deep link to the act. Korea's Ministry of Foreign Affairs hosts the Korean text as a PDF, which redirected without content to our fetcher; per secondary sources the current text is Act No. 21311 (amended Jan 20, 2026).

Cite. Protocol STAMP, “South Korea Framework Act on AI promulgated,” The Docket, entry kr-ai-basic-act-promulgated-2025-01-21, last checked Oct 08, 2026, https://protocolstamp.com/docket/#kr-ai-basic-act-promulgated-2025-01-21.

Egypt National Artificial Intelligence Strategy, second edition (2025-2030)

Issued by the National Council for Artificial Intelligence with a statement from the President of the Republic. Six pillars: governance, technology, data, infrastructure, ecosystem and talent. Plans include a national Arabic large language model and work toward an AI law, building on the Egyptian Charter for Responsible AI.

What it asks to be measured. Governance pillar: responsible-AI framework and a path to binding rules (planned).

Source: National Council for Artificial Intelligence (Egypt). Checked Oct 09, 2026.

Limits: Entry date taken from the official file name (16-1-2025); the pillar list and planned initiatives are from the document's summary pages.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Egypt National Artificial Intelligence Strategy, second edition (2025-2030),” The Docket, entry eg-ai-strategy-2-2025-01-16, last checked Oct 09, 2026, https://protocolstamp.com/docket/#eg-ai-strategy-2-2025-01-16.

Brazil AI bill PL 2338/2023 passes the Senate

The Federal Senate approved a risk-based AI framework bill on Dec 10, 2024. As of October 2026 it is pending in the Chamber of Deputies.

What it asks to be measured. Algorithmic impact assessments for high-risk systems; rights of affected people; a national AI regulation and governance system (proposed).

Source: Federal Senate of Brazil. Checked Oct 08, 2026.

Limits: Chamber status conflicts across secondary sources; check the Chamber page before relying on it.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Brazil AI bill PL 2338/2023 passes the Senate,” The Docket, entry br-pl-2338-senate-2024-12-10, last checked Oct 08, 2026, https://protocolstamp.com/docket/#br-pl-2338-senate-2024-12-10.

International Network of AI Safety Institutes convenes in San Francisco

Founding members (Australia, Canada, the EU, France, Japan, Kenya, Korea, Singapore, the UK and the US) held the network's first meeting and released a joint testing exercise and a mission statement.

What it asks to be measured. Joint testing of models across institutes; shared evaluation methods.

Source: US National Institute of Standards and Technology. Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “International Network of AI Safety Institutes convenes in San Francisco,” The Docket, entry ms-aisi-network-2024-11-20, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ms-aisi-network-2024-11-20.

US and UK AI safety institutes publish their first joint pre-deployment evaluation (upgraded Claude 3.5 Sonnet)

The US AI Safety Institute and UK AI Safety Institute jointly tested Anthropic's upgraded Claude 3.5 Sonnet before release, across biological capabilities, cyber capabilities, software and AI development, and safeguard efficacy, and published the results.

What it asks to be measured. Government pre-deployment testing in four domains, benchmarked against reference models.

Source: US National Institute of Standards and Technology · UK AI Security Institute. Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “US and UK AI safety institutes publish their first joint pre-deployment evaluation (upgraded Claude 3.5 Sonnet),” The Docket, entry intl-aisi-joint-eval-claude-2024-11-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#intl-aisi-joint-eval-claude-2024-11-19.

Product Liability Directive (EU) 2024/2853 covers software and AI systems

Replaces the 1985 directive. Software, including AI systems, is a 'product'; manufacturers can be liable for defects including those arising after updates or machine learning. Member states must transpose by Dec 09, 2026.

What it asks to be measured. Disclosure of evidence by defendants; defectiveness judged with regard to safety requirements including cybersecurity.

Source: EUR-Lex (Official Journal). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Product Liability Directive (EU) 2024/2853 covers software and AI systems,” The Docket, entry eu-pld-2024-11-18, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-pld-2024-11-18.

UK Ministry of Defence JSP 936: Dependable AI in Defence (Part 1: Directive)

The Ministry of Defence's principal policy for adopting AI safely and responsibly. It turns the department's AI ethical principles into direction on governance, development and assurance across the AI lifecycle, and has each part of the MOD name a Responsible AI Senior Officer. It follows the 2022 Defence AI Strategy and the 'Ambitious, Safe, Responsible' policy; the first Responsible AI Senior Officers' report followed in 2025.

What it asks to be measured. Assurance of AI systems across their lifecycle; governance and development requirements; a Responsible AI Senior Officer in each MOD component, reporting on progress.

Source: UK Ministry of Defence · UK Ministry of Defence (directive text) · UK Ministry of Defence (2025 report). Checked Oct 09, 2026.

Limits: The date is taken from the upload time encoded in the published PDF's address; the publication page was not retrieved. Sources were found by web search on Oct 09, 2026 and could not be opened from our fetcher; the summary rests on the search results and their quoted text, not a full read.

Cite. Protocol STAMP, “UK Ministry of Defence JSP 936: Dependable AI in Defence (Part 1: Directive),” The Docket, entry uk-mod-jsp936-2024-11-14, last checked Oct 09, 2026, https://protocolstamp.com/docket/#uk-mod-jsp936-2024-11-14.

Canadian Artificial Intelligence Safety Institute launched

Launched with C$50 million over five years to study AI risks and work with the international network of AI safety institutes.

What it asks to be measured. Research on AI evaluation and safety risks.

Source: Government of Canada. Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “Canadian Artificial Intelligence Safety Institute launched,” The Docket, entry ca-caisi-2024-11-12, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ca-caisi-2024-11-12.

National Security Memorandum on AI

Designated the US AI Safety Institute as industry's primary point of contact for voluntary pre-deployment testing of frontier models for national-security risks, and set a framework for AI governance in national security.

What it asks to be measured. Classified and unclassified pre-deployment testing of frontier models for cyber, biological, chemical and nuclear risks.

Source: The White House (archived). Checked Oct 08, 2026.

Limits: Status under the current administration not confirmed here.

Cite. Protocol STAMP, “National Security Memorandum on AI,” The Docket, entry us-fed-nsm-ai-2024-10-24, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-nsm-ai-2024-10-24.

Montevideo Declaration on a regional approach to AI governance (Latin America and the Caribbean)

Approved at the second Ministerial Summit on the Ethics of AI in Latin America and the Caribbean in Montevideo, Oct 03, 2024 and Oct 04, 2024. Set a 2024-2025 regional roadmap for ethical AI, a standing working group led by Uruguay, and annual ministerial summits.

What it asks to be measured. None directly.

Source: AGESIC, Government of Uruguay. Checked Oct 09, 2026.

Limits: Summit dates and the working-group handover are from the declaration text as reported; check the declaration PDF for exact wording.

Cite. Protocol STAMP, “Montevideo Declaration on a regional approach to AI governance (Latin America and the Caribbean),” The Docket, entry intl-montevideo-declaration-2024-10-04, last checked Oct 09, 2026, https://protocolstamp.com/docket/#intl-montevideo-declaration-2024-10-04.

California SB 1047 (frontier model safety) vetoed

Governor Newsom vetoed the Safe and Secure Innovation for Frontier AI Models Act, which would have required safety testing, a shutdown capability and third-party audits for large models. He convened a working group whose 2025 report informed SB 53.

What it asks to be measured. Would have required pre-deployment safety testing and annual third-party audits (not enacted).

Source: Office of the Governor of California · California Legislature. Checked Oct 08, 2026.

Cite. Protocol STAMP, “California SB 1047 (frontier model safety) vetoed,” The Docket, entry us-ca-sb1047-veto-2024-09-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-sb1047-veto-2024-09-29.

FTC 'Operation AI Comply' against deceptive AI claims

Five law-enforcement actions announced together, including against DoNotPay ('robot lawyer') and businesses selling AI-powered income schemes, for deceptive claims about AI.

What it asks to be measured. Substantiation of claims about what AI products can do.

Source: US Federal Trade Commission. Checked Oct 08, 2026.

Cite. Protocol STAMP, “FTC 'Operation AI Comply' against deceptive AI claims,” The Docket, entry en-us-ftc-ai-comply-2024-09-25, last checked Oct 08, 2026, https://protocolstamp.com/docket/#en-us-ftc-ai-comply-2024-09-25.

Malaysia: National Guidelines on AI Governance and Ethics (AIGE)

Launched by the Ministry of Science, Technology and Innovation to support the National AI Roadmap 2021-2025. Voluntary guidelines for end users, policymakers, designers, developers and suppliers, setting seven principles including fairness, reliability, safety and control, transparency and accountability.

What it asks to be measured. Voluntary principles for responsible AI; no binding evaluation duties.

Source: Government of Malaysia portal. Checked Oct 09, 2026.

Limits: Launch date (Sep 20, 2024) is from secondary reports; the portal page does not print it.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Malaysia: National Guidelines on AI Governance and Ethics (AIGE),” The Docket, entry my-aige-2024-09-20, last checked Oct 09, 2026, https://protocolstamp.com/docket/#my-aige-2024-09-20.

California AI Transparency Act (SB 942) and training-data transparency (AB 2013)

SB 942 covers generative AI systems with over 1,000,000 monthly users: a free AI detection tool, latent disclosures in generated content, and $5,000 per violation. AB 853 (2025) delayed it to Aug 02, 2026 and added platform (2027) and capture-device (2028) duties. AB 2013 requires training-data documentation by Jan 01, 2026 for systems released on or after Jan 01, 2022.

What it asks to be measured. Provenance disclosures and public detection tools (SB 942); published training-data documentation (AB 2013).

Source: California Legislature · California Legislature · California Legislature. Checked Oct 08, 2026.

See explainer: Labelling AI-generated content: when the duty applies.

Cite. Protocol STAMP, “California AI Transparency Act (SB 942) and training-data transparency (AB 2013),” The Docket, entry us-ca-sb942-ab2013-2024-09-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ca-sb942-ab2013-2024-09-19.

China AI Safety Governance Framework 1.0 (TC260)

Released by TC260. Classifies inherent and application risks of AI and sets technical and governance responses. Version 2.0 followed in September 2025.

What it asks to be measured. Risk classification; safety testing and evaluation; traceability and labelling.

Source: TC260 (National Information Security Standardization Technical Committee). Checked Oct 08, 2026.

Cite. Protocol STAMP, “China AI Safety Governance Framework 1.0 (TC260),” The Docket, entry cn-ai-safety-framework-2024-09-09, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-ai-safety-framework-2024-09-09.

Council of Europe Framework Convention on AI (CETS No. 225) opens for signature

Adopted by the Committee of Ministers on May 17, 2024 and opened for signature in Vilnius on Sep 05, 2024; first signatories included the EU, UK and US. First legally binding international treaty on AI, human rights, democracy and the rule of law. Enters into force after five ratifications including three Council of Europe members.

What it asks to be measured. Risk and impact assessment of AI systems, documentation, transparency and oversight measures; remedies for people affected.

Source: Council of Europe · Council of Europe Treaty Office · Council of Europe (Newsroom). Checked Oct 08, 2026.

Limits: Not in force as of Oct 08, 2026 (no entry-into-force date on the Treaty Office page); ratification count changes.

Cite. Protocol STAMP, “Council of Europe Framework Convention on AI (CETS No. 225) opens for signature,” The Docket, entry coe-framework-convention-2024-09-05, last checked Oct 08, 2026, https://protocolstamp.com/docket/#coe-framework-convention-2024-09-05.

Australia Voluntary AI Safety Standard and proposals for mandatory guardrails

Published Sep 05, 2024 by the Department of Industry, Science and Resources: 10 voluntary guardrails, alongside a proposals paper on mandatory guardrails for high-risk AI. On Oct 21, 2025 the National AI Centre published Guidance for AI Adoption (6 essential practices). The mandatory proposal was set aside in the December 2025 National AI Plan.

What it asks to be measured. Testing and monitoring of AI systems; record-keeping; conformity assessments (proposed for high-risk use).

Source: Department of Industry, Science and Resources (Australia). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Australia Voluntary AI Safety Standard and proposals for mandatory guardrails,” The Docket, entry au-voluntary-safety-standard-2024-09-05, last checked Oct 08, 2026, https://protocolstamp.com/docket/#au-voluntary-safety-standard-2024-09-05.

US AI Safety Institute agreements with OpenAI and Anthropic for pre-release model testing

NIST's AI Safety Institute signed memoranda of understanding giving it access to major new models from OpenAI and Anthropic before and after public release, for research, testing and evaluation.

What it asks to be measured. Government pre-deployment and post-deployment evaluation of frontier models.

Source: US National Institute of Standards and Technology. Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “US AI Safety Institute agreements with OpenAI and Anthropic for pre-release model testing,” The Docket, entry us-fed-aisi-agreements-2024-08-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-aisi-agreements-2024-08-29.

Illinois HB 3773: AI in employment under the Human Rights Act

Public Act 103-0804, approved Aug 09, 2024. Amends the Illinois Human Rights Act, effective Jan 01, 2026, to make it a civil-rights violation for employers to use AI that has a discriminatory effect, or zip codes as a proxy, and requires notice to employees when AI is used.

What it asks to be measured. Notice when AI is used in employment decisions; discriminatory-effect standard.

Source: Illinois General Assembly · Illinois General Assembly (Public Act 103-0804). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Illinois HB 3773: AI in employment under the Human Rights Act,” The Docket, entry us-il-hb3773-2024-08-09, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-il-hb3773-2024-08-09.

EU AI Act (Regulation (EU) 2024/1689) enters into force

Published in the Official Journal on Jul 12, 2024 and in force from Aug 01, 2024. Staggered application: prohibitions and AI literacy from Feb 02, 2025; general-purpose AI model duties from Aug 02, 2025; most other provisions from Aug 02, 2026, with high-risk dates later moved by the 2026 Omnibus.

What it asks to be measured. Conformity assessment, risk management, data governance, testing, accuracy and robustness for high-risk AI; model evaluation and adversarial testing for GPAI models with systemic risk; serious-incident reporting; transparency for generated content.

Source: EUR-Lex (Official Journal) · European Commission. Checked Oct 08, 2026.

See explainer: The EU AI Act, from proposal to full application, Where countries stand.

Cite. Protocol STAMP, “EU AI Act (Regulation (EU) 2024/1689) enters into force,” The Docket, entry eu-ai-act-2024-08-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-act-2024-08-01.

Nigeria publishes its draft National Artificial Intelligence Strategy

The Federal Ministry of Communications, Innovation and Digital Economy published a draft strategy, co-created at a workshop in April 2024 and hosted by NITDA's National Centre for AI and Robotics. Five pillars cover infrastructure, the ecosystem, sector adoption, responsible and ethical AI, and governance.

What it asks to be measured. Recommends testing, evaluation and certification infrastructure, bias audits, model documentation and incident reporting (proposed).

Source: NCAIR / NITDA (Nigeria) · Federal Ministry of Communications, Innovation and Digital Economy (Nigeria). Checked Oct 09, 2026.

Limits: Entry dated from the file name (01082024); whether a final text superseded the draft was not confirmed. Evaluation measures listed are from secondary summaries.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Nigeria publishes its draft National Artificial Intelligence Strategy,” The Docket, entry ng-ai-strategy-draft-2024-08-01, last checked Oct 09, 2026, https://protocolstamp.com/docket/#ng-ai-strategy-draft-2024-08-01.

NIST AI 600-1: Generative AI Profile of the AI RMF

Companion profile to the AI RMF identifying 12 risks unique to or exacerbated by generative AI and over 200 suggested actions.

What it asks to be measured. Pre-deployment testing, red-teaming, content provenance, incident disclosure for generative AI.

Source: US National Institute of Standards and Technology · US National Institute of Standards and Technology. Checked Oct 08, 2026.

Cite. Protocol STAMP, “NIST AI 600-1: Generative AI Profile of the AI RMF,” The Docket, entry us-fed-nist-600-1-2024-07-26, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-nist-600-1-2024-07-26.

African Union Continental Artificial Intelligence Strategy endorsed

Endorsed by the AU Executive Council in July 2024. Sets a continental, development-oriented approach to AI, including governance and regulation, for AU member states.

What it asks to be measured. Calls for national AI governance frameworks and risk-based regulation; no specific evaluation duties.

Source: African Union. Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “African Union Continental Artificial Intelligence Strategy endorsed,” The Docket, entry african-union-ai-strategy-2024-07-18, last checked Oct 08, 2026, https://protocolstamp.com/docket/#african-union-ai-strategy-2024-07-18.

NATO publishes a summary of its revised AI strategy

The revision updates NATO's 2021 AI Strategy to speed safe and responsible use of AI across the Alliance. It prioritises putting NATO's six Principles of Responsible Use into practice (lawfulness; responsibility and accountability; explainability and traceability; reliability; governability; bias mitigation), interoperability and work with industry and academia, and names AI-enabled disinformation and information operations as concerns.

What it asks to be measured. Implementation of the Principles of Responsible Use, including reliability and traceability of AI systems. No public test or pass mark.

Source: NATO · GlobalSecurity.org (secondary). Checked Oct 09, 2026.

Limits: Sources were found by web search on Oct 09, 2026 and could not be opened from our fetcher; the summary rests on the search results and their quoted text, not a full read.

Cite. Protocol STAMP, “NATO publishes a summary of its revised AI strategy,” The Docket, entry intl-nato-ai-strategy-rev-2024-07-10, last checked Oct 09, 2026, https://protocolstamp.com/docket/#intl-nato-ai-strategy-rev-2024-07-10.

UAE Cabinet approves the Charter for the Development and Use of Artificial Intelligence

Approved at a Cabinet meeting chaired by the Prime Minister on Jun 10, 2024. The Charter comprises 12 principles on human values and fair, safe treatment, covering among others safety, algorithmic bias, data privacy, transparency, human oversight and compliance with treaties and applicable laws.

What it asks to be measured. Principles on safety, algorithmic bias, transparency and human oversight (non-binding).

Source: UAE Cabinet · UAE Government portal (u.ae). Checked Oct 09, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “UAE Cabinet approves the Charter for the Development and Use of Artificial Intelligence,” The Docket, entry ae-ai-charter-2024-06-10, last checked Oct 09, 2026, https://protocolstamp.com/docket/#ae-ai-charter-2024-06-10.

Singapore Model AI Governance Framework for Generative AI

Nine dimensions for a trusted generative AI ecosystem, from accountability to testing and assurance and content provenance.

What it asks to be measured. Third-party testing and assurance; incident reporting; content provenance.

Source: AI Verify Foundation (secondary). Checked Oct 08, 2026.

Limits: Sources are copies or reports, not the issuer's own site, so the entry is Likely.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Singapore Model AI Governance Framework for Generative AI,” The Docket, entry sg-genai-framework-2024-05-30, last checked Oct 08, 2026, https://protocolstamp.com/docket/#sg-genai-framework-2024-05-30.

Frontier AI Safety Commitments (AI Seoul Summit)

Sixteen AI companies committed to publish safety frameworks setting out how they assess and manage severe risks from frontier models, including thresholds at which risks would be deemed intolerable.

What it asks to be measured. Published frontier safety frameworks; pre-deployment risk assessment; internal and external red-teaming; thresholds and the actions taken if they are crossed.

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Frontier AI Safety Commitments (AI Seoul Summit),” The Docket, entry intl-seoul-commitments-2024-05-21, last checked Oct 08, 2026, https://protocolstamp.com/docket/#intl-seoul-commitments-2024-05-21.

Council of the EU gives final approval to the AI Act

The final legislative step: the Council approved Parliament's first-reading position on May 21, 2024, adopting the AI Act. It was signed on Jun 13, 2024 and published in the Official Journal on Jul 12, 2024.

What it asks to be measured. As in the AI Act entry.

Source: EUR-Lex (procedure history) · Council of the European Union. Checked Oct 08, 2026.

Limits: The Council press release served a bot check on Oct 08, 2026; dates from the EUR-Lex procedure record.

See explainer: The EU AI Act, from proposal to full application.

Cite. Protocol STAMP, “Council of the EU gives final approval to the AI Act,” The Docket, entry eu-ai-act-council-adoption-2024-05-21, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-act-council-adoption-2024-05-21.

Colorado SB24-205: Colorado AI Act (consumer protections for high-risk AI)

First comprehensive US state law on high-risk AI systems used in consequential decisions. Its start date was delayed to Jun 30, 2026 by SB25B-004 (August 2025), then it was repealed and replaced by SB26-189 in May 2026.

What it asks to be measured. Impact assessments, risk-management programmes, notices and algorithmic-discrimination duties for developers and deployers (as enacted in 2024).

Source: Colorado General Assembly · Colorado General Assembly. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Colorado SB24-205: Colorado AI Act (consumer protections for high-risk AI),” The Docket, entry us-co-sb24-205-2024-05-17, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-co-sb24-205-2024-05-17.

Google DeepMind publishes its Frontier Safety Framework

Defined 'Critical Capability Levels' in domains such as autonomy, biosecurity, cybersecurity and machine-learning R&D, with early-warning evaluations to detect when a model approaches them. Later versions followed in 2025.

What it asks to be measured. Early-warning evaluations against Critical Capability Levels; mitigation plans when thresholds are reached.

Source: Google DeepMind. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Google DeepMind publishes its Frontier Safety Framework,” The Docket, entry ms-gdm-fsf-2024-05-17, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ms-gdm-fsf-2024-05-17.

OECD AI Principles revised (Recommendation of the Council on AI)

OECD ministers updated the 2019 Recommendation on AI to address general-purpose and generative AI, information integrity, and the definition of an 'AI system'. The OECD definition was the basis for the EU AI Act's definition.

What it asks to be measured. Definition of an AI system; transparency and responsible disclosure; robustness, security and safety throughout the lifecycle; traceability.

Source: OECD Legal Instruments · OECD.AI (secondary). Checked Oct 08, 2026.

Cite. Protocol STAMP, “OECD AI Principles revised (Recommendation of the Council on AI),” The Docket, entry oecd-ai-principles-rev-2024-05-03, last checked Oct 08, 2026, https://protocolstamp.com/docket/#oecd-ai-principles-rev-2024-05-03.

Japan AI Guidelines for Business (version 1.0)

METI and MIC consolidated earlier guidelines into one voluntary document for AI developers, providers and business users. Later versions followed.

What it asks to be measured. Risk assessment, verification and documentation across the AI lifecycle; transparency to stakeholders.

Source: Ministry of Economy, Trade and Industry (Japan) · Ministry of Economy, Trade and Industry (Japan). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Japan AI Guidelines for Business (version 1.0),” The Docket, entry jp-ai-guidelines-business-2024-04-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#jp-ai-guidelines-business-2024-04-19.

OMB Memorandum M-24-10 on agency use of AI (rescinded 2025)

Required federal agencies to designate Chief AI Officers and apply minimum risk-management practices to 'rights-impacting' and 'safety-impacting' AI. Replaced by M-25-21 in April 2025.

What it asks to be measured. AI impact assessments, real-world testing, independent evaluation and ongoing monitoring for rights- and safety-impacting AI.

Source: Office of Management and Budget (archived). Checked Oct 08, 2026.

Cite. Protocol STAMP, “OMB Memorandum M-24-10 on agency use of AI (rescinded 2025),” The Docket, entry us-fed-omb-m24-10-2024-03-28, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-omb-m24-10-2024-03-28.

UN General Assembly resolution 78/265 on safe, secure and trustworthy AI

First UN General Assembly resolution on AI, adopted without a vote. Encourages member states to develop regulatory and governance approaches for safe, secure and trustworthy AI systems.

What it asks to be measured. Encourages risk monitoring, impact assessment, testing and evaluation across the AI lifecycle; content authentication and provenance mechanisms.

Source: United Nations Digital Library · United Nations (Meetings Coverage). Checked Oct 08, 2026.

Cite. Protocol STAMP, “UN General Assembly resolution 78/265 on safe, secure and trustworthy AI,” The Docket, entry un-ga-res-78-265-2024-03-21, last checked Oct 08, 2026, https://protocolstamp.com/docket/#un-ga-res-78-265-2024-03-21.

India MeitY advisory on AI models and deepfakes (revised)

MeitY advisory of Mar 01, 2024, revised Mar 15, 2024, asking intermediaries to label AI-generated content and inform users of unreliability of under-tested models; the revision dropped a requirement for prior government permission.

What it asks to be measured. Labelling of potentially unreliable outputs; metadata or identifiers for synthetic content.

Source: Ministry of Electronics and Information Technology. Checked Oct 08, 2026.

Limits: The advisory text was not located at a stable MeitY URL; summary from contemporaneous reporting. Keep as Likely unless the PDF is found.

Cite. Protocol STAMP, “India MeitY advisory on AI models and deepfakes (revised),” The Docket, entry in-meity-advisory-2024-03-15, last checked Oct 08, 2026, https://protocolstamp.com/docket/#in-meity-advisory-2024-03-15.

Utah Artificial Intelligence Policy Act (SB 149)

Signed March 2024, effective May 01, 2024. Requires disclosure when consumers interact with generative AI in certain circumstances and created the Office of Artificial Intelligence Policy and a regulatory learning laboratory.

What it asks to be measured. Disclosure of generative AI interaction; regulatory mitigation agreements for testing in a learning lab.

Source: Utah State Legislature. Checked Oct 08, 2026.

Limits: Signing date (Mar 13, 2024) from the Legislature page; the May 01, 2024 effective date is Utah's default for the 2024 session and is not printed on that page. Later amendments (2025) not checked here.

Cite. Protocol STAMP, “Utah Artificial Intelligence Policy Act (SB 149),” The Docket, entry us-ut-sb149-2024-03-13, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-ut-sb149-2024-03-13.

European Parliament adopts the AI Act (523-46-49)

Plenary approved the AI Act text agreed with the Council in December 2023, by 523 votes to 46 with 49 abstentions. Council adoption followed on May 21, 2024.

What it asks to be measured. As in the AI Act entry.

Source: European Parliament. Checked Oct 08, 2026.

See explainer: The EU AI Act, from proposal to full application.

Cite. Protocol STAMP, “European Parliament adopts the AI Act (523-46-49),” The Docket, entry eu-ai-act-plenary-2024-03-13, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-act-plenary-2024-03-13.

China TC260-003: Basic security requirements for generative AI services

Technical document from National Information Security Standardization Technical Committee (TC260) setting security requirements for training data, models and safety measures; used in the CAC filing process. Later developed into national standards in 2025.

What it asks to be measured. Sampling-based assessment of training data; refusal-rate and answer-safety test question banks; keyword filtering.

Source: TC260 (National Information Security Standardization Technical Committee). Checked Oct 08, 2026.

Limits: TC260's site serves a script page to automated fetchers; the PDF was not opened. URL from a policy archive's record of the original link.

Cite. Protocol STAMP, “China TC260-003: Basic security requirements for generative AI services,” The Docket, entry cn-tc260-genai-security-2024-02-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-tc260-genai-security-2024-02-29.

Russia Presidential Decree No. 124 updates the National AI Strategy to 2030

Amends Decree No. 490 of Oct 10, 2019 and the National Strategy for the Development of AI to 2030. Adds definitions including large generative models, large foundation models and trusted AI technologies, directs federal bodies to follow the strategy in sectoral planning, and requires annual reporting to the President.

What it asks to be measured. Defines 'trusted' AI technologies; reporting on strategy implementation.

Source: Official Internet Portal of Legal Information (Russia) · President of Russia (Decree No. 490 of 2019, as amended). Checked Oct 09, 2026.

Limits: Definitions and reporting duty are from the official summary of the amended text; the full consolidated text was not read.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Russia Presidential Decree No. 124 updates the National AI Strategy to 2030,” The Docket, entry ru-decree124-2024-02-15, last checked Oct 09, 2026, https://protocolstamp.com/docket/#ru-decree124-2024-02-15.

Japan AI Safety Institute (J-AISI) established

Set up within the Information-technology Promotion Agency (IPA). Published a Guide to Evaluation Perspectives on AI Safety and a Guide to Red Teaming Methodology in 2024.

What it asks to be measured. AI safety evaluation perspectives; red-teaming methodology.

Source: Japan AI Safety Institute. Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “Japan AI Safety Institute (J-AISI) established,” The Docket, entry jp-aisi-2024-02-14, last checked Oct 08, 2026, https://protocolstamp.com/docket/#jp-aisi-2024-02-14.

FCC rules AI-generated voices in robocalls are 'artificial' under the TCPA

Declaratory Ruling FCC 24-17 confirmed that calls using AI-generated voice clones fall under the Telephone Consumer Protection Act's restrictions on artificial or prerecorded voices.

What it asks to be measured. Defines AI voice cloning as an 'artificial' voice for consent and disclosure requirements.

Source: US Federal Communications Commission. Checked Oct 08, 2026.

Cite. Protocol STAMP, “FCC rules AI-generated voices in robocalls are 'artificial' under the TCPA,” The Docket, entry us-fed-fcc-ai-voices-2024-02-08, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-fcc-ai-voices-2024-02-08.

UK government response to the AI white paper consultation

Confirmed the sector-regulator approach, asked key regulators to publish AI strategies by Apr 30, 2024, and set out initial thinking on future binding requirements for developers of highly capable general-purpose AI.

What it asks to be measured. Regulators' AI strategies; possible future binding duties on frontier developers (not enacted).

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

Cite. Protocol STAMP, “UK government response to the AI white paper consultation,” The Docket, entry uk-white-paper-response-2024-02-06, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-white-paper-response-2024-02-06.

Commission Decision establishing the European AI Office

Commission Decision C(2024) 390 created the AI Office within DG CONNECT to implement and enforce the AI Act's rules on general-purpose AI models.

What it asks to be measured. Evaluation of general-purpose AI models, including models with systemic risk; monitoring and enforcement.

Source: European Commission · European Commission. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Commission Decision establishing the European AI Office,” The Docket, entry eu-ai-office-2024-01-24, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-office-2024-01-24.

FTC bans Rite Aid from using facial recognition for five years

Settlement over Rite Aid's use of facial-recognition surveillance in stores that the FTC said falsely flagged consumers, disproportionately women and people of colour. The FTC alleged Rite Aid failed to test, assess, measure or document the system's accuracy before deploying it, and that false positives were more frequent in stores in plurality-Black and Asian communities.

What it asks to be measured. Pre-deployment testing and assessment of accuracy (the failure alleged); the order's conditions on any future use were not read here.

Source: US Federal Trade Commission. Checked Oct 08, 2026.

Cite. Protocol STAMP, “FTC bans Rite Aid from using facial recognition for five years,” The Docket, entry en-us-ftc-rite-aid-2023-12-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#en-us-ftc-rite-aid-2023-12-19.

Indonesia: Ministerial Circular No. 9 of 2023 on AI Ethics

The Minister of Communication and Informatics issued an ethics circular for businesses programming with AI and for public and private electronic system operators. AI is not to be used as the decision-maker on matters of humanity; developers must provide information to prevent harm and apply risk and crisis management.

What it asks to be measured. Risk management in AI development; information duties on developers.

Source: JDIH, Ministry of Communication and Digital Affairs (Indonesia). Checked Oct 09, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Indonesia: Ministerial Circular No. 9 of 2023 on AI Ethics,” The Docket, entry id-se9-2023-12-19, last checked Oct 09, 2026, https://protocolstamp.com/docket/#id-se9-2023-12-19.

ISO/IEC 42001:2023 AI management system standard published

First certifiable international management-system standard for AI. Specifies requirements for establishing, maintaining and improving an AI management system within an organisation; third-party certification is available.

What it asks to be measured. AI risk assessment and AI system impact assessment; internal audit; third-party certification audits.

Source: International Organization for Standardization. Checked Oct 08, 2026.

Limits: A voluntary standard; the text is paywalled.

Cite. Protocol STAMP, “ISO/IEC 42001:2023 AI management system standard published,” The Docket, entry iso-iec-42001-2023-12-18, last checked Oct 08, 2026, https://protocolstamp.com/docket/#iso-iec-42001-2023-12-18.

OpenAI publishes its Preparedness Framework (beta)

Set out how OpenAI tracks, evaluates and forecasts catastrophic risks from frontier models, with risk categories and scorecards, and a rule that only models with post-mitigation scores of 'medium' or below can be deployed. Updated in April 2025.

What it asks to be measured. Capability evaluations across tracked risk categories (e.g. cybersecurity, CBRN, persuasion, model autonomy); scorecards; deployment thresholds.

Source: OpenAI · OpenAI (PDF). Checked Oct 08, 2026.

Cite. Protocol STAMP, “OpenAI publishes its Preparedness Framework (beta),” The Docket, entry ms-openai-preparedness-2023-12-18, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ms-openai-preparedness-2023-12-18.

EU institutions reach political agreement on the AI Act

Parliament and Council negotiators agreed the AI Act on Dec 08, 2023, including rules for general-purpose AI models and a list of banned applications.

What it asks to be measured. Model evaluations, adversarial testing and serious-incident reporting for high-impact GPAI models (as agreed).

Source: European Parliament. Checked Oct 08, 2026.

See explainer: The EU AI Act, from proposal to full application.

Cite. Protocol STAMP, “EU institutions reach political agreement on the AI Act,” The Docket, entry eu-ai-act-political-deal-2023-12-08, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-act-political-deal-2023-12-08.

Israel publishes its first policy on AI regulation and ethics

The Ministry of Innovation, Science and Technology, with the Ministry of Justice, recommended sector-specific regulation rather than a horizontal AI law, a risk-based approach consistent with leading countries, 'soft' regulatory tools such as sandboxes, and a coordination centre for AI policy.

What it asks to be measured. Risk-based, sector-by-sector regulation (recommended).

Source: Ministry of Innovation, Science and Technology (Israel). Checked Oct 09, 2026.

Limits: The issuer dates this document to the month only; the entry is placed on the first day of that month.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Israel publishes its first policy on AI regulation and ethics,” The Docket, entry il-ai-policy-2023-12-01, last checked Oct 09, 2026, https://protocolstamp.com/docket/#il-ai-policy-2023-12-01.

UK AI Safety Institute established

Announced at the Bletchley summit as the successor to the Frontier AI Taskforce, with a mission to evaluate advanced AI systems. Renamed the AI Security Institute on Feb 14, 2025.

What it asks to be measured. Pre- and post-deployment evaluations of advanced AI models; development of evaluation tools (e.g. the open-source Inspect framework, May 2024).

Source: UK Government (GOV.UK) · AI Security Institute. Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “UK AI Safety Institute established,” The Docket, entry uk-aisi-2023-11-02, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-aisi-2023-11-02.

Bletchley Declaration (AI Safety Summit, UK)

Signed by 28 countries and the EU at the first AI Safety Summit. Recognises risks from frontier AI and states that developers of highly capable systems bear responsibility for their safety, including through safety testing.

What it asks to be measured. Safety testing and evaluation of frontier AI by developers; transparency on plans to measure, monitor and mitigate dangerous capabilities.

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Bletchley Declaration (AI Safety Summit, UK),” The Docket, entry intl-bletchley-2023-11-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#intl-bletchley-2023-11-01.

Vice President Harris's London speech on the future of AI announces a US AI Safety Institute

Speaking at the US Embassy in London on the eve of the Bletchley summit, the Vice President announced the US AI Safety Institute and argued that AI safety must cover the full spectrum of AI risk, including harms to individuals and institutions as well as existential threats.

What it asks to be measured. Standards for testing the safety of AI models for public use (to be set by the new institute).

“create rigorous standards to test the safety of AI models for public use”Kamala Harris, Vice President of the United States, official White House transcript (archived)

Source: The White House (archived). Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “Vice President Harris's London speech on the future of AI announces a US AI Safety Institute,” The Docket, entry us-fed-harris-ai-speech-2023-11-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-harris-ai-speech-2023-11-01.

G7 Hiroshima Process: Guiding Principles and Code of Conduct for advanced AI systems

G7 leaders endorsed 11 guiding principles and a voluntary International Code of Conduct for organisations developing advanced AI, including foundation models and generative AI.

What it asks to be measured. Evaluation across the AI lifecycle including red-teaming; public reporting of capabilities, limitations and domains of appropriate use; incident information-sharing; content authentication and provenance.

Source: European Commission · G7 Hiroshima AI Process (Ministry of Internal Affairs and Communications, Japan). Checked Oct 08, 2026.

Cite. Protocol STAMP, “G7 Hiroshima Process: Guiding Principles and Code of Conduct for advanced AI systems,” The Docket, entry g7-hiroshima-code-2023-10-30, last checked Oct 08, 2026, https://protocolstamp.com/docket/#g7-hiroshima-code-2023-10-30.

US Executive Order 14110 on safe, secure and trustworthy AI (revoked 2025)

Required developers of dual-use foundation models above a compute threshold (10^26 operations) to report red-team results to the government under the Defense Production Act, and directed NIST guidance on evaluations. Revoked on Jan 20, 2025 by EO 14148.

What it asks to be measured. Reporting of red-team test results for large models; NIST red-teaming and evaluation guidance; watermarking guidance.

Source: Federal Register. Checked Oct 08, 2026.

See explainer: US federal AI policy, 2020 to 2026, Training-compute thresholds that trigger AI rules.

Cite. Protocol STAMP, “US Executive Order 14110 on safe, secure and trustworthy AI (revoked 2025),” The Docket, entry us-fed-eo14110-2023-10-30, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-eo14110-2023-10-30.

Prime Minister Sunak's speech on AI announces a UK AI Safety Institute

At the Royal Society, a week before the Bletchley summit, the Prime Minister announced the UK AI Safety Institute to evaluate and test new AI models, and named loss of control of AI as an extreme risk, referring to 'super intelligence'.

What it asks to be measured. Government testing and evaluation of new frontier models by a national institute.

“carefully examine, evaluate, and test new types of AI”Rishi Sunak, Prime Minister of the United Kingdom, official GOV.UK transcript

“there is even the risk that humanity could lose control of AI completely”Rishi Sunak, Prime Minister of the United Kingdom, official GOV.UK transcript

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

See explainer: Government AI testing bodies.

Cite. Protocol STAMP, “Prime Minister Sunak's speech on AI announces a UK AI Safety Institute,” The Docket, entry uk-sunak-ai-speech-2023-10-26, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-sunak-ai-speech-2023-10-26.

Santiago Declaration to promote ethical AI in Latin America and the Caribbean

Adopted at the first Ministerial and High Authorities Summit of Latin America and the Caribbean in Santiago de Chile, Oct 23, 2023 and Oct 24, 2023, organised by Chile's Ministry of Science with UNESCO and CAF. Agreed a working group toward an intergovernmental AI council for the region, building on the UNESCO Recommendation.

What it asks to be measured. None directly; regional commitment to ethical-impact frameworks.

Source: Ministerio de Ciencia, Tecnología, Conocimiento e Innovación (Chile). Checked Oct 09, 2026.

Limits: Signatory count (about 20) is from press reports.

Cite. Protocol STAMP, “Santiago Declaration to promote ethical AI in Latin America and the Caribbean,” The Docket, entry intl-santiago-declaration-2023-10-24, last checked Oct 09, 2026, https://protocolstamp.com/docket/#intl-santiago-declaration-2023-10-24.

DSA Delegated Regulation (EU) 2024/436 on independent audits

Adopted by the Commission on Oct 20, 2023. Sets rules for how independent auditors audit very large online platforms under Art. 37 DSA, including methodology, audit risk and report templates.

What it asks to be measured. Audit methodology, auditor independence, report and implementation templates for audits that include algorithmic systems.

Source: EUR-Lex (Official Journal). Checked Oct 08, 2026.

Cite. Protocol STAMP, “DSA Delegated Regulation (EU) 2024/436 on independent audits,” The Docket, entry eu-dsa-audit-delegated-2023-10-20, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-dsa-audit-delegated-2023-10-20.

China announces the Global AI Governance Initiative

Announced at the Belt and Road Forum. Calls for a people-centred approach, a testing and assessment system based on AI risk levels, and a greater role for developing countries in global AI governance.

What it asks to be measured. A risk-level-based testing and assessment system (proposed).

Source: Ministry of Foreign Affairs of the PRC. Checked Oct 08, 2026.

Cite. Protocol STAMP, “China announces the Global AI Governance Initiative,” The Docket, entry cn-global-ai-governance-initiative-2023-10-18, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-global-ai-governance-initiative-2023-10-18.

Canada: Voluntary Code of Conduct on advanced generative AI systems

ISED code with commitments on accountability, safety, fairness, transparency, human oversight and robustness; signatories added over time.

What it asks to be measured. Adversarial testing and red-teaming; third-party audits for public-facing systems; detection methods for generated content.

Source: Innovation, Science and Economic Development Canada. Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Canada: Voluntary Code of Conduct on advanced generative AI systems,” The Docket, entry ca-genai-code-2023-09-27, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ca-genai-code-2023-09-27.

Anthropic publishes its Responsible Scaling Policy

Introduced 'AI Safety Levels' (ASL) with capability thresholds that trigger stronger safety and security measures, and commitments to evaluate models for dangerous capabilities before scaling further. Revised in later versions.

What it asks to be measured. Dangerous-capability evaluations at defined points in training and before deployment; ASL thresholds.

Source: Anthropic. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Anthropic publishes its Responsible Scaling Policy,” The Docket, entry ms-anthropic-rsp-2023-09-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ms-anthropic-rsp-2023-09-19.

China: Measures for Ethical Review of Science and Technology Activities (trial)

Issued by the Ministry of Science and Technology and other agencies; effective Dec 01, 2023. Requires ethics review committees and lists activities needing expert re-review, including some algorithm and automated decision-making systems with public-opinion mobilisation capacity.

What it asks to be measured. Ethics review, and expert re-review for listed high-risk activities.

Source: Ministry of Science and Technology of the PRC. Checked Oct 08, 2026.

Cite. Protocol STAMP, “China: Measures for Ethical Review of Science and Technology Activities (trial),” The Docket, entry cn-ethics-review-measures-2023-09-07, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-ethics-review-measures-2023-09-07.

Saudi Arabia: SDAIA AI Ethics Principles (version 1.0)

The Saudi Data and AI Authority issued AI Ethics Principles applying to all AI stakeholders in the Kingdom across the AI lifecycle: fairness, privacy and security, humanity, social and environmental benefits, reliability and safety, transparency and explainability, and accountability and responsibility. Generative AI guidelines for government and the public followed.

What it asks to be measured. Risk categorisation of AI systems; controls per principle across the lifecycle; self-assessment.

Source: Saudi Data and AI Authority (SDAIA). Checked Oct 09, 2026.

Limits: The issuer dates this document to the month only; the entry is placed on the first day of that month.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Saudi Arabia: SDAIA AI Ethics Principles (version 1.0),” The Docket, entry sa-sdaia-ai-ethics-2023-09-01, last checked Oct 09, 2026, https://protocolstamp.com/docket/#sa-sdaia-ai-ethics-2023-09-01.

White House voluntary AI commitments (seven companies)

Amazon, Anthropic, Google, Inflection, Meta, Microsoft and OpenAI committed to safety, security and trust measures; eight more companies joined in September 2023.

What it asks to be measured. Internal and external security testing before release, carried out in part by independent experts; information-sharing on risks; watermarking or provenance for AI audio and visual content; public reporting of capabilities and limitations.

Source: The White House (archived). Checked Oct 08, 2026.

Cite. Protocol STAMP, “White House voluntary AI commitments (seven companies),” The Docket, entry us-fed-voluntary-commitments-2023-07-21, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-voluntary-commitments-2023-07-21.

China: Interim Measures for the Management of Generative AI Services

Issued by the CAC and six other agencies; effective Aug 15, 2023. Applies to generative AI services offered to the public in China.

What it asks to be measured. Security assessments and algorithm filing for services with public-opinion attributes; training-data legality and quality; labelling of generated content.

Source: Cyberspace Administration of China. Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “China: Interim Measures for the Management of Generative AI Services,” The Docket, entry cn-genai-measures-2023-07-10, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-genai-measures-2023-07-10.

Peru Law 31814 promoting the use of AI, and its regulation (DS 115-2025-PCM)

Law 31814 (July 2023) was among the first AI laws in Latin America. Its implementing regulation, Supreme Decree 115-2025-PCM, was published on Sep 09, 2025 with risk-based provisions for public entities and voluntary adoption by the private sector.

What it asks to be measured. Risk classification of AI uses; transparency and human oversight for higher-risk uses by public entities.

Source: Presidency of the Council of Ministers (Peru). Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Peru Law 31814 promoting the use of AI, and its regulation (DS 115-2025-PCM),” The Docket, entry pe-law-31814-2023-07-05, last checked Oct 08, 2026, https://protocolstamp.com/docket/#pe-law-31814-2023-07-05.

Argentina approves Recommendations for Trustworthy AI in the public sector (Disposición 2/2023)

The Undersecretariat of Information Technologies approved 'Recommendations for a Trustworthy Artificial Intelligence', published in the Official Bulletin on Jun 02, 2023, for public innovation projects using AI. Principles include proportionality, safety, fairness, transparency and explainability, and accountability with oversight and audit mechanisms.

What it asks to be measured. Risk assessment across the AI project lifecycle; audit and oversight mechanisms; informing people when AI is used.

Source: Argentina.gob.ar (official legal database). Checked Oct 09, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Argentina approves Recommendations for Trustworthy AI in the public sector (Disposición 2/2023),” The Docket, entry ar-disposicion-2-2023-06-02, last checked Oct 09, 2026, https://protocolstamp.com/docket/#ar-disposicion-2-2023-06-02.

FTC, CFPB, DOJ and EEOC joint statement on automated systems

Four federal enforcement agencies stated that existing civil-rights, consumer-protection and fair-competition laws apply to automated systems and AI.

What it asks to be measured. None new; existing law applies to outcomes of automated systems.

Source: US Federal Trade Commission · US Federal Trade Commission (PDF). Checked Oct 08, 2026.

Cite. Protocol STAMP, “FTC, CFPB, DOJ and EEOC joint statement on automated systems,” The Docket, entry us-fed-joint-statement-2023-04-25, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-joint-statement-2023-04-25.

Rwanda National Artificial Intelligence Policy approved

Approved by Cabinet in April 2023; developed by the Ministry of ICT and Innovation with the regulator RURA. Aims to use AI for growth while positioning Rwanda as a responsible-AI champion; priorities include AI literacy, infrastructure, an open data ecosystem, public-sector adoption and practical ethical guidelines. A readiness and maturity framework tracks implementation.

What it asks to be measured. Practical AI ethics guidelines; a readiness and maturity framework to track progress.

Source: Ministry of ICT and Innovation (Rwanda) · Ministry of ICT and Innovation (Rwanda). Checked Oct 09, 2026.

Limits: The Cabinet date (Apr 20, 2023) is from secondary reports of the Prime Minister's Office communique, not from the policy document.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Rwanda National Artificial Intelligence Policy approved,” The Docket, entry rw-ai-policy-2023-04-20, last checked Oct 09, 2026, https://protocolstamp.com/docket/#rw-ai-policy-2023-04-20.

UK AI regulation white paper: a pro-innovation approach

Proposed a non-statutory, principles-based framework applied by existing sector regulators rather than a single AI law. Five principles: safety, security and robustness; transparency and explainability; fairness; accountability and governance; contestability and redress.

What it asks to be measured. Regulators to apply principles within their remits; assurance techniques and technical standards as tools.

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

See explainer: Where countries stand.

Cite. Protocol STAMP, “UK AI regulation white paper: a pro-innovation approach,” The Docket, entry uk-ai-white-paper-2023-03-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-ai-white-paper-2023-03-29.

GPT-4 released with a technical report and system card

OpenAI released GPT-4 with a technical report that withheld architecture, model size and training details, and a system card describing pre-deployment red-teaming by external experts, including an evaluation by the Alignment Research Center of risky emergent behaviours.

What it asks to be measured. External red-teaming before release; published system card; dangerous-capability evaluation by a third party (ARC).

Source: arXiv (OpenAI, GPT-4 Technical Report) · OpenAI (system card PDF). Checked Oct 08, 2026.

Cite. Protocol STAMP, “GPT-4 released with a technical report and system card,” The Docket, entry ms-gpt4-system-card-2023-03-14, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ms-gpt4-system-card-2023-03-14.

Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy

A non-binding declaration launched by the United States at the first REAIM summit sets out measures for military use of AI, including compliance with international humanitarian law and human responsibility. 31 states had endorsed it by November 2023 and about 50 by 2024; China, Russia, India, Israel, Brazil and Saudi Arabia had not, as of March 2024. Endorsing states meet each year.

What it asks to be measured. Testing and assurance of military AI capabilities and human responsibility for their use; the text was not retrieved for this entry.

Source: US Department of State · Fox News (secondary) · Lawfare (secondary). Checked Oct 09, 2026.

Limits: Endorsement counts differ by source and date (45, 49, 53 and 54 were all reported in 2024); no 2026 count was found. The launch date is from contemporaneous accounts of the 2023 REAIM summit. Sources were found by web search on Oct 09, 2026 and could not be opened from our fetcher; the summary rests on the search results and their quoted text, not a full read.

Cite. Protocol STAMP, “Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy,” The Docket, entry intl-us-political-declaration-2023-02-16, last checked Oct 09, 2026, https://protocolstamp.com/docket/#intl-us-political-declaration-2023-02-16.

NIST AI Risk Management Framework 1.0 (NIST AI 100-1)

Voluntary framework organised around four functions (Govern, Map, Measure, Manage). Widely referenced by later laws, including Colorado's and Texas's AI acts as a compliance reference.

What it asks to be measured. The Measure function: test, evaluation, verification and validation (TEVV) of AI risks and trustworthiness characteristics.

Source: US National Institute of Standards and Technology · US National Institute of Standards and Technology. Checked Oct 08, 2026.

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “NIST AI Risk Management Framework 1.0 (NIST AI 100-1),” The Docket, entry us-fed-nist-ai-rmf-2023-01-26, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-nist-ai-rmf-2023-01-26.

US DoD Directive 3000.09 updated: Autonomy in Weapon Systems

The Department of Defense updated its policy for autonomous and semi-autonomous weapon systems. Systems must be designed to let commanders and operators exercise appropriate levels of human judgment over the use of force, and most need senior-level review before formal development and again before fielding. A national security memorandum of Jun 05, 2026 (NSPM-11) ordered the directive rewritten.

What it asks to be measured. Verification and validation, and test and evaluation, of autonomous weapon systems before fielding; senior review before development and before fielding. The department has not published how many systems it has reviewed.

Source: US Department of Defense (directive text) · Congressional Research Service (secondary) · Wikipedia (secondary). Checked Oct 09, 2026.

Limits: The directive text was not retrieved; the summary rests on the Congressional Research Service and other secondary accounts. Whether the rewrite ordered by NSPM-11 within 90 days has been issued was not confirmed. Sources were found by web search on Oct 09, 2026 and could not be opened from our fetcher; the summary rests on the search results and their quoted text, not a full read.

Cite. Protocol STAMP, “US DoD Directive 3000.09 updated: Autonomy in Weapon Systems,” The Docket, entry us-fed-dodd-3000-09-2023-01-25, last checked Oct 09, 2026, https://protocolstamp.com/docket/#us-fed-dodd-3000-09-2023-01-25.

China: Provisions on Administration of Deep Synthesis Internet Information Services

Issued by the CAC, MIIT and MPS; effective Jan 10, 2023. Applies to deepfake and synthetic-media services.

What it asks to be measured. Labelling of deep-synthesis content; security assessments; algorithm filing; real-name verification.

Source: Cyberspace Administration of China. Checked Oct 08, 2026.

See explainer: Labelling AI-generated content: when the duty applies.

Cite. Protocol STAMP, “China: Provisions on Administration of Deep Synthesis Internet Information Services,” The Docket, entry cn-deep-synthesis-2022-11-25, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-deep-synthesis-2022-11-25.

Digital Services Act (Regulation (EU) 2022/2065): audits of very large platforms' algorithmic systems

Adopted Oct 19, 2022; applied to very large online platforms and search engines from August 2023 and to all services from Feb 17, 2024. Very large platforms must assess systemic risks, including from recommender and algorithmic systems, and undergo annual independent audits.

What it asks to be measured. Annual systemic-risk assessments covering algorithmic systems; annual independent audits (Art. 37); researcher data access; recommender-system transparency.

Source: EUR-Lex (Official Journal) · European Commission. Checked Oct 08, 2026.

Limits: Not AI-specific; included because it mandates independent audits of algorithmic systems.

Cite. Protocol STAMP, “Digital Services Act (Regulation (EU) 2022/2065): audits of very large platforms' algorithmic systems,” The Docket, entry eu-dsa-2022-10-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-dsa-2022-10-19.

Blueprint for an AI Bill of Rights (White House OSTP)

Non-binding framework of five principles for automated systems: safe and effective systems; algorithmic discrimination protections; data privacy; notice and explanation; human alternatives and fallback.

What it asks to be measured. Pre-deployment testing, risk identification and ongoing monitoring; independent evaluation; plain-language reporting.

Source: The White House (archived) · White House OSTP (archived). Checked Oct 08, 2026.

Limits: Withdrawn from the live White House site in January 2025; the archive remains.

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “Blueprint for an AI Bill of Rights (White House OSTP),” The Docket, entry us-fed-ai-bill-of-rights-2022-10-04, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-ai-bill-of-rights-2022-10-04.

AI Liability Directive proposed (later withdrawn)

Proposal to ease claims for damage caused by AI, including disclosure of evidence on high-risk AI systems and a rebuttable presumption of causality. Listed for withdrawal in the Commission's 2025 work programme in February 2025.

What it asks to be measured. Court-ordered disclosure of documentation and logs for high-risk AI systems (proposed, not adopted).

Source: EUR-Lex. Checked Oct 08, 2026.

Limits: Withdrawal status from the Commission work programme; confirm against the Official Journal notice.

Cite. Protocol STAMP, “AI Liability Directive proposed (later withdrawn),” The Docket, entry eu-ai-liability-proposal-2022-09-28, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-liability-proposal-2022-09-28.

Canada Bill C-27 / Artificial Intelligence and Data Act (died on the Order Paper)

Tabled Jun 16, 2022. AIDA would have regulated 'high-impact' AI systems. The bill died when Parliament was prorogued on Jan 06, 2025.

What it asks to be measured. Would have required risk assessment, mitigation, monitoring and record-keeping for high-impact systems (not enacted).

Source: Parliament of Canada (LEGISinfo). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Canada Bill C-27 / Artificial Intelligence and Data Act (died on the Order Paper),” The Docket, entry ca-c27-aida-2022-06-16, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ca-c27-aida-2022-06-16.

Singapore launches AI Verify testing framework and toolkit

IMDA and PDPC launched AI Verify as a pilot governance testing framework and software toolkit; open-sourced in 2023 under the AI Verify Foundation.

What it asks to be measured. Technical tests and process checks against 11 governance principles, mapped to international frameworks.

Source: Infocomm Media Development Authority · AI Verify Foundation (secondary). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Singapore launches AI Verify testing framework and toolkit,” The Docket, entry sg-ai-verify-2022-05-25, last checked Oct 08, 2026, https://protocolstamp.com/docket/#sg-ai-verify-2022-05-25.

China: Provisions on Algorithmic Recommendation Management for Internet Information Services

Issued by the CAC and three other agencies; effective Mar 01, 2022. Created the algorithm filing (registry) system.

What it asks to be measured. Algorithm filing with the CAC; security self-assessments; user options to switch off personalised recommendation; algorithm transparency.

Source: Cyberspace Administration of China. Checked Oct 08, 2026.

Cite. Protocol STAMP, “China: Provisions on Algorithmic Recommendation Management for Internet Information Services,” The Docket, entry cn-algorithm-recommendation-2021-12-31, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-algorithm-recommendation-2021-12-31.

New York City Local Law 144: bias audits of automated employment decision tools

Enacted December 2021; enforcement began Jul 05, 2023 after DCWP rules. Employers and agencies using an automated employment decision tool must have an independent bias audit within the past year and publish a summary.

What it asks to be measured. Annual independent bias audit with impact ratios by sex and race/ethnicity; public summary; notice to candidates.

Source: New York City Council · NYC Department of Consumer and Worker Protection. Checked Oct 08, 2026.

Limits: A city law; jurisdiction coded US-NY.

Cite. Protocol STAMP, “New York City Local Law 144: bias audits of automated employment decision tools,” The Docket, entry us-nyc-ll144-2021-12-11, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-nyc-ll144-2021-12-11.

CDEI roadmap to an effective AI assurance ecosystem

Centre for Data Ethics and Innovation roadmap describing how audit, certification, impact assessment and other assurance services for AI could develop in the UK.

What it asks to be measured. Types of AI assurance: impact assessment, bias audit, compliance audit, certification, conformity assessment, performance testing.

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

Cite. Protocol STAMP, “CDEI roadmap to an effective AI assurance ecosystem,” The Docket, entry uk-cdei-assurance-roadmap-2021-12-08, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-cdei-assurance-roadmap-2021-12-08.

Algorithmic Transparency Recording Standard (ATRS)

Published by the Cabinet Office and CDEI in November 2021. Made mandatory for central government departments in February 2024 (per the government's white paper response).

What it asks to be measured. Public records of algorithmic tools used in public-sector decisions: purpose, data, human oversight, risks.

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

Cite. Protocol STAMP, “Algorithmic Transparency Recording Standard (ATRS),” The Docket, entry uk-atrs-2021-11-29, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-atrs-2021-11-29.

UNESCO Recommendation on the Ethics of Artificial Intelligence

Adopted by UNESCO's 193 member states at the 41st General Conference. The first global standard-setting instrument on AI ethics; it asks states to set up ethical impact assessment and readiness assessment tools and report on implementation.

What it asks to be measured. Ethical impact assessments of AI systems; national readiness assessments (UNESCO's Readiness Assessment Methodology); periodic state reporting.

Source: UNESCO · UNESCO (UNESDOC). Checked Oct 08, 2026.

Cite. Protocol STAMP, “UNESCO Recommendation on the Ethics of Artificial Intelligence,” The Docket, entry intl-unesco-ethics-2021-11-23, last checked Oct 08, 2026, https://protocolstamp.com/docket/#intl-unesco-ethics-2021-11-23.

China: Ethical Norms for New Generation Artificial Intelligence

Issued by the National New Generation AI Governance Specialist Committee under the Ministry of Science and Technology. Six basic ethical requirements and specific norms for management, R&D, supply and use.

What it asks to be measured. Ethical review and risk assessment across the AI lifecycle; controllability and human decision-making authority.

Source: Ministry of Science and Technology of the PRC. Checked Oct 08, 2026.

Cite. Protocol STAMP, “China: Ethical Norms for New Generation Artificial Intelligence,” The Docket, entry cn-ethics-norms-2021-09-25, last checked Oct 08, 2026, https://protocolstamp.com/docket/#cn-ethics-norms-2021-09-25.

UK National AI Strategy

Published Sep 22, 2021. Committed to developing an AI assurance ecosystem and to a white paper on AI governance and regulation.

What it asks to be measured. Commitment to an AI assurance roadmap and AI standards hub.

Source: UK Government (GOV.UK). Checked Oct 08, 2026.

Cite. Protocol STAMP, “UK National AI Strategy,” The Docket, entry uk-national-ai-strategy-2021-09-22, last checked Oct 08, 2026, https://protocolstamp.com/docket/#uk-national-ai-strategy-2021-09-22.

Turkey's National Artificial Intelligence Strategy 2021-2025 published in the Official Gazette

Presidential Circular 2021/18 put into force the strategy prepared by the Presidency's Digital Transformation Office and the Ministry of Industry and Technology: six strategic priorities, 24 objectives and 119 measures, including regulation to speed socioeconomic adaptation. A 2024-2025 action plan updated it.

What it asks to be measured. None stated for evaluation.

Source: Resmî Gazete (Official Gazette of Türkiye) · Presidency of the Republic of Türkiye, Digital Transformation Office. Checked Oct 09, 2026.

Limits: Priority, objective and measure counts are from the Digital Transformation Office's summary.

See explainer: Where countries stand.

Cite. Protocol STAMP, “Turkey's National Artificial Intelligence Strategy 2021-2025 published in the Official Gazette,” The Docket, entry tr-ai-strategy-2021-08-20, last checked Oct 09, 2026, https://protocolstamp.com/docket/#tr-ai-strategy-2021-08-20.

Colorado SB21-169: insurers' use of external data and algorithms

Prohibits insurers from unfair discrimination through external consumer data, algorithms and predictive models; the Division of Insurance adopted governance and testing regulations for life insurers in 2023.

What it asks to be measured. Governance frameworks and quantitative testing of algorithms for unfairly discriminatory outcomes.

Source: Colorado General Assembly. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Colorado SB21-169: insurers' use of external data and algorithms,” The Docket, entry us-co-sb21-169-2021-07-06, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-co-sb21-169-2021-07-06.

European Commission proposes the Artificial Intelligence Act (COM(2021) 206)

Proposal for a regulation laying down harmonised rules on AI using a risk-based structure: prohibited practices, high-risk systems subject to conformity assessment, transparency duties for certain systems.

What it asks to be measured. Conformity assessment, quality management, testing, accuracy, robustness and post-market monitoring for high-risk AI.

Source: EUR-Lex. Checked Oct 08, 2026.

See explainer: The EU AI Act, from proposal to full application.

Cite. Protocol STAMP, “European Commission proposes the Artificial Intelligence Act (COM(2021) 206),” The Docket, entry eu-ai-act-proposal-2021-04-21, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-ai-act-proposal-2021-04-21.

FTC guidance: 'Aiming for truth, fairness, and equity in your company's use of AI'

FTC business guidance stating that sale or use of racially biased algorithms may violate Section 5 of the FTC Act, the FCRA or ECOA, and urging companies to test algorithms for discriminatory outcomes.

What it asks to be measured. Testing algorithms for discriminatory outcomes before and during use; truthful claims about AI.

Source: US Federal Trade Commission (Internet Archive copy). Checked Oct 08, 2026.

Limits: The post now returns 404 at ftc.gov (checked Oct 08, 2026); cited from the Internet Archive capture of Jan 01, 2025.

Cite. Protocol STAMP, “FTC guidance: 'Aiming for truth, fairness, and equity in your company's use of AI',” The Docket, entry us-fed-ftc-ai-guidance-2021-04-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-ftc-ai-guidance-2021-04-19.

African Commission on Human and Peoples' Rights Resolution 473 on AI and human rights

Adopted at the Commission's 31st Extraordinary Session, held virtually Feb 19, 2021 to Feb 25, 2021. Calls on States to keep AI consistent with the African Charter and to ensure meaningful human control over AI systems, and launches a study on AI, robotics and emerging technologies in Africa; a draft study was published in March 2025.

What it asks to be measured. Meaningful human control over consequential AI systems.

Source: African Commission on Human and Peoples' Rights. Checked Oct 09, 2026.

Limits: Entry dated to the last day of the session; the page does not print the day of adoption.

Cite. Protocol STAMP, “African Commission on Human and Peoples' Rights Resolution 473 on AI and human rights,” The Docket, entry intl-achpr-res473-2021-02-25, last checked Oct 09, 2026, https://protocolstamp.com/docket/#intl-achpr-res473-2021-02-25.

National Artificial Intelligence Initiative Act of 2020

Enacted as Division E of the FY2021 National Defense Authorization Act (veto overridden Jan 01, 2021). Created the National AI Initiative Office and directed NIST to develop a voluntary AI risk management framework.

What it asks to be measured. Directed NIST to develop the AI Risk Management Framework and AI measurement standards.

Source: US Congress (Congress.gov). Checked Oct 08, 2026.

Limits: The AI provisions are within the NDAA (Public Law 116-283).

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “National Artificial Intelligence Initiative Act of 2020,” The Docket, entry us-fed-niaia-2021-01-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-niaia-2021-01-01.

US Executive Order 13960: trustworthy AI in the federal government

Set nine principles for federal agencies' use of AI and required agencies to inventory their AI use cases.

What it asks to be measured. Annual agency AI use-case inventories; AI that is accurate, reliable, effective, safe, transparent, accountable and regularly monitored.

Source: Federal Register · US Government Publishing Office. Checked Oct 08, 2026.

See explainer: US federal AI policy, 2020 to 2026.

Cite. Protocol STAMP, “US Executive Order 13960: trustworthy AI in the federal government,” The Docket, entry us-fed-eo13960-2020-12-03, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-fed-eo13960-2020-12-03.

Assessment List for Trustworthy AI (ALTAI) published

Final self-assessment checklist from the Commission's High-Level Expert Group on AI, turning the 2019 Ethics Guidelines' seven requirements into an operational tool for organisations.

What it asks to be measured. Self-assessment against seven requirements: human agency and oversight, technical robustness and safety, privacy, transparency, diversity and fairness, societal well-being, accountability.

Source: European Commission. Checked Oct 08, 2026.

Limits: Voluntary self-assessment tool.

Cite. Protocol STAMP, “Assessment List for Trustworthy AI (ALTAI) published,” The Docket, entry eu-altai-2020-07-17, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-altai-2020-07-17.

Russia Federal Law No. 123-FZ: experimental legal regime for AI in Moscow

Signed Apr 24, 2020. Set up, from Jul 01, 2020 and for five years, an experimental legal regime for developing and deploying AI technologies in Moscow, with a register of participants, and amended the federal Personal Data law to allow processing of depersonalised data within the regime. The law defines 'artificial intelligence' and 'AI technologies'.

What it asks to be measured. None stated; defines AI in federal law and sets conditions for participants in the regime.

Source: Official Internet Portal of Legal Information (Russia) · President of Russia (kremlin.ru). Checked Oct 09, 2026.

Limits: Whether the regime was extended after its five-year term was not checked.

Cite. Protocol STAMP, “Russia Federal Law No. 123-FZ: experimental legal regime for AI in Moscow,” The Docket, entry ru-123fz-2020-04-24, last checked Oct 09, 2026, https://protocolstamp.com/docket/#ru-123fz-2020-04-24.

Canada: Directive on Automated Decision-Making compliance date

Treasury Board directive (issued 2019) with which federal departments had to comply from Apr 01, 2020. Requires an Algorithmic Impact Assessment before production use of an automated decision system; updated in 2023 and 2025.

What it asks to be measured. Algorithmic Impact Assessment (published); peer review for higher-impact levels; testing for unintended bias; human intervention.

Source: Treasury Board of Canada Secretariat · Government of Canada. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Canada: Directive on Automated Decision-Making compliance date,” The Docket, entry ca-directive-adm-2020-04-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#ca-directive-adm-2020-04-01.

European Commission White Paper on Artificial Intelligence

Set out policy options for an EU regulatory framework for AI, proposing mandatory requirements for 'high-risk' applications and prior conformity assessment. Precursor to the AI Act.

What it asks to be measured. Proposed prior conformity assessment, testing and certification for high-risk AI; record-keeping and transparency.

Source: European Commission · European Commission (PDF). Checked Oct 08, 2026.

Cite. Protocol STAMP, “European Commission White Paper on Artificial Intelligence,” The Docket, entry eu-white-paper-ai-2020-02-19, last checked Oct 08, 2026, https://protocolstamp.com/docket/#eu-white-paper-ai-2020-02-19.

Singapore Model AI Governance Framework, second edition

PDPC and IMDA updated the voluntary framework launched at Davos 2019, with an Implementation and Self-Assessment Guide for Organisations.

What it asks to be measured. Self-assessment; human involvement levels by risk; explainability, repeatability and robustness testing.

Source: Personal Data Protection Commission Singapore. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Singapore Model AI Governance Framework, second edition,” The Docket, entry sg-model-framework-2020-01-21, last checked Oct 08, 2026, https://protocolstamp.com/docket/#sg-model-framework-2020-01-21.

Illinois Artificial Intelligence Video Interview Act takes effect

Employers using AI to analyse video interviews must notify applicants, explain how the AI works, and obtain consent; 2022 amendments added demographic reporting where AI alone decides who gets an in-person interview.

What it asks to be measured. Notice and consent; annual demographic data reporting to the state (from 2022).

Source: Illinois General Assembly. Checked Oct 08, 2026.

Cite. Protocol STAMP, “Illinois Artificial Intelligence Video Interview Act takes effect,” The Docket, entry us-il-aivia-2020-01-01, last checked Oct 08, 2026, https://protocolstamp.com/docket/#us-il-aivia-2020-01-01.