{"citation_id":"TSIO-D03-20261108","title":"SI control map: laws, rules and frameworks","description":"43 laws, regulations, executive actions and frameworks that govern frontier SI in 22 jurisdictions and bodies, with status, key dates, scope, duties, penalties and sources, plus the safety policies of frontier labs.","version":"2026.11.08","publisher":"The SI Observatory (tsio.si)","license":"CC BY 4.0","license_url":"https://creativecommons.org/licenses/by/4.0/","dataset_page":"https://tsio.si/data/si-control-map/","version_url":"https://tsio.si/data/si-control-map/v2026.11.08/","sources":["Official texts and regulators, linked per measure in source_urls.","The SI Observatory, https://tsio.si/report/#rules"],"columns":{"measures":[{"name":"jurisdiction_id","type":"text","description":"Jurisdiction id."},{"name":"jurisdiction","type":"text","description":"Jurisdiction or body."},{"name":"measure_id","type":"text","description":"Measure id."},{"name":"short_name","type":"text","description":"Short name."},{"name":"full_name","type":"text","description":"Official name."},{"name":"status","type":"text","description":"in-force, passed, proposed, voluntary or stalled."},{"name":"introduced","type":"date","description":"Date introduced."},{"name":"signed","type":"date","description":"Date signed or adopted."},{"name":"effective","type":"date","description":"Date it took or takes effect."},{"name":"next_date","type":"date","description":"Next milestone date."},{"name":"next_milestone","type":"text","description":"What happens then."},{"name":"scope","type":"text","description":"Who and what it covers."},{"name":"requires","type":"text","description":"Main duties, separated by \" | \"."},{"name":"penalties","type":"text","description":"Penalties."},{"name":"source_urls","type":"text","description":"Sources, separated by \" | \"."}],"lab_policies":[{"name":"id","type":"text","description":"Policy id."},{"name":"lab","type":"text","description":"The lab."},{"name":"policy","type":"text","description":"Policy name."},{"name":"version","type":"text","description":"Version."},{"name":"updated","type":"date","description":"Last update."},{"name":"summary","type":"text","description":"What it says."},{"name":"url","type":"url","description":"The policy."}]},"tables":{"measures":[{"jurisdiction_id":"european-union","jurisdiction":"European Union","measure_id":"eu-ai-act-gpai","short_name":"EU AI Act (GPAI)","full_name":"Regulation (EU) 2024/1689 (Artificial Intelligence Act), Chapter V: obligations for providers of general-purpose AI models","status":"in-force","introduced":null,"signed":"2024-06-13","effective":"2025-08-02","next_date":"2027-08-02","next_milestone":"Deadline for general purpose SI models placed on the market before August 2, 2025 to comply","scope":"Providers placing general purpose SI models on the EU market, with added duties for models with systemic risk, presumed above 10^25 FLOP of training compute.","requires":"Notify the Commission within two weeks of reaching or foreseeing the 10^25 FLOP systemic risk threshold | Keep technical documentation and give information to downstream providers | Publish a summary of training content and maintain a copyright policy | Systemic risk models: evaluate the model, including adversarial testing, and assess and mitigate systemic risks | Systemic risk models: track and report serious incidents to the AI Office | Systemic risk models: ensure adequate cybersecurity for the model and its infrastructure","penalties":"From August 2, 2026 the Commission can fine providers of general purpose SI models up to 3 percent of worldwide annual turnover or 15 million euros, whichever is higher.","source_urls":"https://digital-strategy.ec.europa.eu/en/policies/guidelines-gpai-providers | https://artificialintelligenceact.eu/gpai-guidelines-overview/ | https://www.reedsmith.com/our-insights/blogs/technology-law-dispatch/102o6as/eu-ai-act-the-ai-office-sends-its-first-requests-for-information-to-leading-ai/"},{"jurisdiction_id":"european-union","jurisdiction":"European Union","measure_id":"eu-gpai-code-safety-security","short_name":"GPAI Code of Practice (Safety and Security)","full_name":"General-Purpose AI Code of Practice: Safety and Security Chapter","status":"voluntary","introduced":null,"signed":"2025-07-10","effective":"2025-08-02","next_date":null,"next_milestone":null,"scope":"Voluntary for providers of general purpose SI models with systemic risk; the Commission and AI Board treat it as an adequate tool for showing AI Act compliance, and signatories include Amazon, Anthropic, Google, IBM, Microsoft, Mistral AI and OpenAI, with xAI signing only this chapter and Meta not signing.","requires":"Adopt a Safety and Security Framework and notify it to the AI Office | Identify and analyze systemic risks, including CBRN, loss of control, cyber offense and harmful manipulation | Set measurable risk acceptance criteria and proceed only when all risks are acceptable | Use independent external evaluators, with limited exemptions | Track and report serious incidents and file a Safety and Security Model Report | Secure the model and maintain whistleblower protections and a healthy risk culture","penalties":"","source_urls":"https://digital-strategy.ec.europa.eu/en/policies/gpai-code-practice | https://digital-strategy.ec.europa.eu/en/faqs/signing-general-purpose-ai-code-practice | https://en.wikipedia.org/wiki/General-Purpose_AI_Code_of_Practice"},{"jurisdiction_id":"european-union","jurisdiction":"European Union","measure_id":"eu-digital-omnibus-ai","short_name":"AI Omnibus","full_name":"Regulation (EU) 2026/1744 (Digital Omnibus on AI)","status":"in-force","introduced":null,"signed":"2026-07-08","effective":"2026-07-27","next_date":"2026-12-02","next_milestone":"Generative systems already on the market must meet the Article 50(2) marking duty, and added Article 5 provisions apply","scope":"Amends the AI Act and gives the AI Office direct supervision of SI systems built on a general purpose model by the same provider and of SI systems in very large online platforms and search engines.","requires":"Delays stand-alone Annex III high-risk obligations to December 2, 2027 | Delays high-risk obligations for Annex I product components to August 2, 2028 | Gives the AI Office information, inspection, binding commitment and fining powers (Articles 75a to 75c) | Gives generative systems placed on the market before August 2, 2026 until December 2, 2026 to meet output marking duties | Does not delay the general purpose model obligations or their August 2, 2026 enforcement start","penalties":"Allows periodic penalty payments capped at 5 percent of average daily worldwide turnover per day, alongside existing AI Act fines.","source_urls":"https://artificialintelligenceact.eu/ai-act-explorer/digital-omnibus/ | https://www.whitecase.com/insight-alert/eu-ai-omnibus-enters-force-amending-ai-act"},{"jurisdiction_id":"us","jurisdiction":"United States (federal)","measure_id":"us-eo-14434","short_name":"EO 14434","full_name":"Executive Order 14434, Inaugurating the Era of Super Intelligence","status":"in-force","introduced":null,"signed":"2026-09-29","effective":"2026-09-29","next_date":"2026-11-28","next_milestone":"Assistant to the President for Science and Technology must submit proposed legislative language defining Super Intelligence (60 days after signing)","scope":"Federal executive agencies; it places no duties on private SI developers.","requires":"Agencies use the terms \"Super Intelligence\" and \"SI\" in place of the older terms in official communications and non-statutory documents, to the extent permitted by law | Defines Super Intelligence as the technologies covered by the existing statutory definition in 15 U.S.C. 9401(3) until superseded | Within 60 days, propose legislative language on whether the new definition should modify or supersede the statutory one, with conforming amendments | Recommend any additional executive action | Existing regulations, contracts, grants and historical documents need not be changed","penalties":"","source_urls":"https://www.presidency.ucsb.edu/documents/executive-order-14434-inaugurating-the-era-super-intelligence | https://www.whitehouse.gov/wp-content/uploads/2026/09/eo-14434.pdf"},{"jurisdiction_id":"us","jurisdiction":"United States (federal)","measure_id":"us-eo-14365","short_name":"EO 14365","full_name":"Executive Order 14365, Ensuring a National Policy Framework for Artificial Intelligence","status":"in-force","introduced":null,"signed":"2025-12-11","effective":"2025-12-11","next_date":null,"next_milestone":null,"scope":"Federal agencies, directed at state SI laws the administration considers onerous; it does not itself preempt any state law.","requires":"Attorney General forms an AI Litigation Task Force to challenge state SI laws (within 30 days) | Commerce publishes an evaluation of onerous state SI laws (due about March 11, 2026; no public release found as of November 2026) | Condition non-deployment BEAD broadband funds on states not enforcing such laws | FCC to consider a federal reporting and disclosure standard; FTC proposed a related policy statement on July 1, 2026 | Advisers prepare a legislative recommendation for a preemptive federal framework | The United States joined xAI's suit against Colorado's 2024 law as a plaintiff on April 24, 2026","penalties":"","source_urls":"https://www.federalregister.gov/documents/2025/12/16/2025-23092/ensuring-a-national-policy-framework-for-artificial-intelligence | https://www.ailawsbystate.com/blog/federal-ai-preemption-doj-task-force-state-law-fight-2026 | https://casrai.org/news/federal-ai-moratorium-state-preemption-fight-2026"},{"jurisdiction_id":"us","jurisdiction":"United States (federal)","measure_id":"us-ai-action-plan","short_name":"AI Action Plan","full_name":"Winning the Race: America's AI Action Plan","status":"voluntary","introduced":null,"signed":"2025-07-23","effective":null,"next_date":null,"next_milestone":null,"scope":"Federal agencies; more than 90 policy actions across three pillars, with no binding duties on private SI developers.","requires":"Remove federal regulations that hinder SI development and adoption | Speed permits for data centers and semiconductor fabrication plants | Export full stack American SI packages to allies | Contract only with frontier model developers whose systems are objective and free from top-down ideological bias","penalties":"","source_urls":"https://www.whitehouse.gov/releases/2025/07/white-house-unveils-americas-ai-action-plan | https://oecd.ai/en/dashboards/policy-initiatives/americas-ai-action-plan"},{"jurisdiction_id":"us","jurisdiction":"United States (federal)","measure_id":"us-national-policy-framework","short_name":"National Policy Framework","full_name":"National Policy Framework for Artificial Intelligence","status":"voluntary","introduced":null,"signed":"2026-03-20","effective":null,"next_date":null,"next_milestone":null,"scope":"Legislative recommendations from the White House to Congress; it is not binding.","requires":"Broad federal preemption of unduly burdensome state SI laws in favor of one national standard | States should not regulate SI model development, which is treated as interstate commerce | States keep generally applicable child protection, fraud and consumer protection laws | No new federal SI regulator; rely on existing agencies and industry standards | Age assurance and harm reduction features for SI platforms likely to be used by minors","penalties":"","source_urls":"https://wilmerhale.com/en/insights/blogs/wilmerhale-privacy-and-cybersecurity-law/20260323-white-house-releases-national-policy-framework-for-artificial-intelligence"},{"jurisdiction_id":"us","jurisdiction":"United States (federal)","measure_id":"us-caissi","short_name":"CAISSI","full_name":"Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), NIST","status":"voluntary","introduced":null,"signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Voluntary agreements with private SI developers and evaluators; formerly the US AI Safety Institute and then the Center for AI Standards and Innovation (CAISI), renamed CAISSI in 2026.","requires":"Act as industry's main government contact for testing and research on commercial SI systems | Lead unclassified evaluations of SI capabilities that may pose national security risks, focused on cyber, biological and chemical risks | Assess US, adversary and foreign SI systems, including possible backdoors | Develop voluntary guidelines and standards, such as incident readiness practices published October 16, 2026","penalties":"","source_urls":"https://www.nist.gov/caissi | https://tech-insider.org/caissi-ai-safety-center-white-house-reporting-independence-2026"},{"jurisdiction_id":"us","jurisdiction":"United States (federal)","measure_id":"us-frontier-act","short_name":"FRONTIER Act","full_name":"Frontier Risk Oversight, National Transparency, Independent Evaluation, and Reporting Act (FRONTIER Act), H.R. 9925","status":"proposed","introduced":"2026-07-23","signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Frontier SI developers, with duties tiered by developer size; it grew out of the Great American AI Act discussion draft released June 4, 2026.","requires":"Publish model cards and risk management frameworks | Undergo independent audits and ongoing assessments | Report safety incidents | Create a uniform national standard for catastrophic risk transparency, auditing and reporting that would displace state rules","penalties":"","source_urls":"https://legiscan.com/US/bill/HB9925/ | https://www.benton.org/headlines/reps-obernolte-trahan-introduce-bipartisan-frontier-act-strengthen-oversight-advanced-ai | https://fpf.org/blog/frontier-ai-goes-federal-how-the-great-american-ai-act-compares-to-state-laws/"},{"jurisdiction_id":"california","jurisdiction":"California","measure_id":"ca-sb-53","short_name":"SB 53","full_name":"Transparency in Frontier Artificial Intelligence Act (SB 53, Chapter 138, Statutes of 2025)","status":"in-force","introduced":"2025-01-07","signed":"2025-09-29","effective":"2026-01-01","next_date":"2027-01-01","next_milestone":"First annual Cal OES report on critical safety incidents, Attorney General whistleblower report and Department of Technology review of frontier definitions due","scope":"Developers that train frontier models above 10^26 operations; large frontier developers, with over $500 million in annual revenue, carry the heaviest duties.","requires":"Large developers publish a frontier SI framework, review it yearly and post material changes within 30 days | Publish a transparency report before deploying a new or substantially modified frontier model | Report critical safety incidents to Cal OES within 15 days, or within 24 hours if there is imminent risk of death or serious injury | Large developers send Cal OES summaries of internal catastrophic risk assessments every three months | Protect whistleblowers, with an anonymous internal reporting channel at large developers","penalties":"Civil penalties of up to $1 million per violation, enforced only by the California Attorney General.","source_urls":"https://calmatters.digitaldemocracy.org/bills/ca_202520260sb53 | https://casrai.org/guides/california-sb-53-frontier-ai-transparency-act | https://thenextweb.com/news/openai-sb-53-california-ai-safety-law-amendments"},{"jurisdiction_id":"california","jurisdiction":"California","measure_id":"ca-sb-813","short_name":"SB 813","full_name":"SB 813, Independent verification organizations (Chapter 179, Statutes of 2026)","status":"passed","introduced":"2025-02-21","signed":"2026-09-09","effective":"2027-01-01","next_date":"2027-05-01","next_milestone":"Executive Order N-9-26 target for posting criteria for independent verification organizations (statutory deadline January 1, 2028)","scope":"SI auditors seeking state designation as independent verification organizations; use of them by developers is voluntary.","requires":"Government Operations Agency sets application, suspension and qualification criteria by January 1, 2028 | Agency convenes working groups that include engineers and SI safety experts | Designated organizations file annual reports with the agency and the Legislature | No one is required to use an organization or be audited; an audit is relevant but not conclusive in harm lawsuits","penalties":"","source_urls":"https://calmatters.digitaldemocracy.org/bills/ca_202520260sb813 | https://www.wilmerhale.com/en/insights/blogs/wilmerhale-privacy-and-cybersecurity-law/20261002-california-finalizes-next-wave-of-state-ai-and-privacy-regulation"},{"jurisdiction_id":"california","jurisdiction":"California","measure_id":"ca-eo-n-9-26","short_name":"EO N-9-26","full_name":"Executive Order N-9-26","status":"in-force","introduced":null,"signed":"2026-09-18","effective":"2026-09-18","next_date":"2026-11-16","next_milestone":"Government Operations Agency recommendations due on onsite audits, a frontier model kill switch and loss of control incident reporting","scope":"California state agencies; the recommendations are not binding and any new duties for frontier SI developers would need legislation.","requires":"Recommend periodic onsite audits by independent verification organizations inside large frontier developers | Recommend independent verification of safety frameworks, transparency reports and risk assessments | Recommend a kill switch for frontier models whose effectiveness is verified on an ongoing basis | Recommend expanding critical safety incident reporting to loss of control incidents | Post criteria for independent verification organizations by May 1, 2027","penalties":"","source_urls":"https://ppc.land/newsom-sets-november-16-deadline-to-study-frontier-ai-kill-switch/ | https://www.radarfirst.com/blog/california-ai-governance-independent-verification/ | https://calmatters.org/politics/2026/09/ai-rules-newsom-state-directive/"},{"jurisdiction_id":"new-york","jurisdiction":"New York","measure_id":"ny-raise-act","short_name":"RAISE Act","full_name":"Responsible AI Safety and Education Act (RAISE Act), S6953B/A6453B as amended by S8828/A9449","status":"passed","introduced":null,"signed":"2025-12-19","effective":"2027-01-01","next_date":"2027-01-01","next_milestone":"Law takes effect and the Department of Financial Services office begins oversight","scope":"Frontier developers training models above 10^26 FLOP; large frontier developers have at least $500 million in prior year revenue, including affiliates.","requires":"Publish a transparency report at or before deploying a new or substantially modified frontier model | Large developers publish and annually review a frontier SI framework | Report critical safety incidents to the DFS office within 72 hours, or 24 hours if there is imminent risk of death or serious injury | Large developers submit quarterly summaries of internal catastrophic risk assessments to DFS | Large developers file a disclosure statement every two years and pay a share of the office's costs","penalties":"Attorney General may seek up to $1 million for a first violation and up to $3 million for later violations, plus $1,000 per day for disclosure filing failures.","source_urls":"https://www.dwt.com/blogs/artificial-intelligence-law-advisor/2026/04/ny-overhauls-frontier-ai-transparency-law | https://privacy-daily.com/news/2026/03/27/NY-Raise-Act-Chapter-Amendments-Signed-by-Gov-Hochul-2603270021 | https://www.dwt.com/blogs/artificial-intelligence-law-advisor/2025/12/new-york-raise-act-ai-safety-rules-developers"},{"jurisdiction_id":"illinois","jurisdiction":"Illinois","measure_id":"il-sb-315","short_name":"SB 315","full_name":"Artificial Intelligence Safety Measures Act (SB 315)","status":"passed","introduced":null,"signed":"2026-07-06","effective":"2027-01-01","next_date":"2027-01-01","next_milestone":"Act takes effect; framework and annual audit duties follow on January 1, 2028","scope":"Frontier developers training models above 10^26 FLOP whose models are accessible in Illinois; large frontier developers exceed $500 million in annual revenue.","requires":"Publish transparency reports before deploying new or substantially modified models | Large developers publish and annually update a safety framework from January 1, 2028 | Large developers commission annual independent third party audits and post a summary within 30 days | Report critical safety incidents to the Illinois Emergency Management Agency and Attorney General within 72 hours, or 24 hours for imminent risk | Protect whistleblowers and provide anonymous internal reporting channels","penalties":"Civil penalties of up to $1 million for a first violation and up to $3 million for later violations, enforced by the Attorney General.","source_urls":"https://gov-pritzker-newsroom.prezly.com/gov-pritzker-signs-nation-leading-artificial-intelligence-safety-law | https://www.crowell.com/en/insights/client-alerts/illinois-imposes-transparency-and-safety-obligations-on-frontier-ai-systems | https://www.hunton.com/privacy-and-cybersecurity-law-blog/illinois-governor-signs-frontier-ai-model-law"},{"jurisdiction_id":"colorado","jurisdiction":"Colorado","measure_id":"co-sb-24-205","short_name":"SB 24-205","full_name":"Colorado Artificial Intelligence Act (SB 24-205)","status":"stalled","introduced":null,"signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Developers and deployers of high-risk SI systems used in consequential decisions.","requires":"Duty of reasonable care against algorithmic discrimination | Deployer risk management programs and annual impact assessments | Start date delayed to June 30, 2026 by SB25B-004, signed August 28, 2025 | Enforcement paused by a stipulated order in X.AI LLC v. Weiser on April 27, 2026, then repealed and replaced by SB 26-189","penalties":"","source_urls":"https://www.pymnts.com/artificial-intelligence-2/2025/colorado-delays-ai-act-to-june-2026-businesses-get-breathing-room/ | https://www.ailawsbystate.com/blog/federal-ai-preemption-doj-task-force-state-law-fight-2026"},{"jurisdiction_id":"colorado","jurisdiction":"Colorado","measure_id":"co-sb-26-189","short_name":"SB 26-189","full_name":"SB 26-189 (repeal and reenactment of the Colorado AI Act)","status":"passed","introduced":null,"signed":"2026-05-14","effective":"2027-01-01","next_date":"2027-01-01","next_milestone":"Law takes effect; Attorney General rules on consumer disclosures due the same day","scope":"Developers and deployers of automated decision-making technology used in consequential decisions; it does not address frontier model safety.","requires":"Developers give deployers documentation on intended uses, known harmful uses and limitations | Deployers notify consumers before use and explain adverse outcomes afterward | Consumers may correct data and request human review where commercially reasonable | Keep records for at least three years","penalties":"Violations are deceptive trade practices enforced only by the Attorney General, with a 60 day cure period until January 1, 2030.","source_urls":"https://www.hklaw.com/en/insights/publications/2026/05/colorado-governor-signs-sb-189 | https://casrai.org/news/federal-ai-moratorium-state-preemption-fight-2026"},{"jurisdiction_id":"texas","jurisdiction":"Texas","measure_id":"tx-traiga","short_name":"TRAIGA","full_name":"Texas Responsible Artificial Intelligence Governance Act (HB 149)","status":"in-force","introduced":null,"signed":"2025-06-22","effective":"2026-01-01","next_date":null,"next_milestone":null,"scope":"Developers and deployers of SI systems in Texas, plus state agencies.","requires":"Bars SI systems built with intent to manipulate behavior, discriminate unlawfully or produce sexual deepfakes of children | State agencies disclose when residents interact with SI | Bans capturing biometric identifiers without consent | Regulatory sandbox under a new Artificial Intelligence Council | 60 day cure period before penalties","penalties":"Attorney General enforcement only, with fines of $80,000 to $200,000 per uncured violation.","source_urls":"https://iapp.org/news/a/governor-signs-texas-responsible-artificial-intelligence-governance-act"},{"jurisdiction_id":"china","jurisdiction":"China","measure_id":"cn-generative-measures","short_name":"Generative AI Measures","full_name":"Interim Measures for the Management of Generative Artificial Intelligence Services","status":"in-force","introduced":null,"signed":"2023-07-10","effective":"2023-08-15","next_date":null,"next_milestone":null,"scope":"Providers of generative SI services to the public in China, including through APIs.","requires":"Security assessment and algorithm filing for services with public opinion or social mobilization capacity | Ensure training data is lawful, respecting intellectual property and personal information law | Mark generated content conspicuously | Prevent, remove and report illegal content","penalties":"Penalized under existing laws such as the Cybersecurity, Data Security and Personal Information Protection laws; no standalone fine.","source_urls":"https://www.twobirds.com/en/insights/2023/china/what-you-need-to-know-about-china’s-new-generative-ai-measures | https://pwccn.com/en/industries/telecommunications-media-and-technology/publications/interim-measures-for-generative-ai-services-implemented-aug2023.html"},{"jurisdiction_id":"china","jurisdiction":"China","measure_id":"cn-labeling-measures","short_name":"Labeling Measures","full_name":"Measures for Labeling of AI-Generated Synthetic Content","status":"in-force","introduced":null,"signed":"2025-03-14","effective":"2025-09-01","next_date":null,"next_milestone":null,"scope":"Online service providers that generate or distribute synthetic text, images, audio, video and virtual scenes, plus app distribution platforms and users.","requires":"Add explicit visible labels to synthetic content | Embed implicit labels in file metadata | App stores check whether apps offer generative services and review labeling materials | Users declare synthetic content when publishing it | Follow the mandatory national labeling standard that took effect the same day","penalties":"Handled by internet, telecom, public security and broadcasting authorities under existing laws.","source_urls":"https://loeb.com/en/insights/publications/2025/03/chinas-ai-labeling-measures-and-mandatory-national-standards-take-effect-september-1"},{"jurisdiction_id":"china","jurisdiction":"China","measure_id":"cn-cybersecurity-law-amendment","short_name":"Cybersecurity Law SI clause","full_name":"Cybersecurity Law of the People's Republic of China, as amended October 28, 2025 (Article 20)","status":"in-force","introduced":null,"signed":"2025-10-28","effective":"2026-01-01","next_date":null,"next_milestone":null,"scope":"National legal basis for state support and oversight of SI development.","requires":"State support for SI research, training data and computing infrastructure | Improve SI ethics norms | Strengthen SI risk monitoring and assessment","penalties":"Fines for particularly serious violations of the Cybersecurity Law rise to RMB 10 million.","source_urls":"https://www.caixinglobal.com/2025-10-29/china-amends-cybersecurity-law-to-add-first-ever-ai-clause-102376988.html | https://www.china-briefing.com/news/china-ai-regulation-2026-new-laws-rules/"},{"jurisdiction_id":"china","jurisdiction":"China","measure_id":"cn-ai-safety-governance-framework","short_name":"Framework 3.0","full_name":"AI Safety Governance Framework 3.0","status":"voluntary","introduced":null,"signed":"2026-09-14","effective":null,"next_date":null,"next_milestone":null,"scope":"Non-binding technical guidance from standards committee TC260, issued under Cyberspace Administration of China guidance, for SI developers and service providers.","requires":"Adds a standalone agentic SI risk category covering permission abuse, planning failures and tool execution | Describes loss of control risks, including models deceiving evaluators and sandbagging | Treats loss of control over chemical, biological, nuclear and missile knowledge as a risk | Adds an agentic risk management framework and trustworthy SI principles | Updates version 2.0, released September 15, 2025","penalties":"","source_urls":"https://obsidianri.com/blog/newsroom-cn-20260915-ai-safety-governance-framework-3 | https://aisafetychina.substack.com/p/brief-29-what-chinas-new-ai-safety | https://sesec.eu/?p=2036"},{"jurisdiction_id":"united-kingdom","jurisdiction":"United Kingdom","measure_id":"uk-ai-security-institute","short_name":"UK AISI","full_name":"AI Security Institute","status":"voluntary","introduced":null,"signed":"2025-02-14","effective":"2025-02-14","next_date":null,"next_milestone":null,"scope":"Voluntary pre-deployment testing and research with frontier SI developers.","requires":"Focus on serious SI risks with security implications, such as chemical and biological weapons and cyberattacks | Criminal misuse team working with the Home Office | Work with defense, cyber security and national security agencies | Joint evaluations with partner institutes, such as a July 23, 2026 cyber assessment with CAISSI","penalties":"","source_urls":"https://www.gov.uk/government/news/tackling-ai-security-risks-to-unleash-growth-and-deliver-plan-for-change | https://www.nist.gov/caissi"},{"jurisdiction_id":"united-kingdom","jurisdiction":"United Kingdom","measure_id":"uk-frontier-bill","short_name":"UK frontier bill","full_name":"Binding regulation for developers of the most powerful AI models (2024 commitment)","status":"stalled","introduced":null,"signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Would apply binding rules to the few companies developing the most powerful SI models.","requires":"Labour's 2024 manifesto promised binding regulation on the handful of companies developing the most powerful models | The 2024 King's Speech made a tentative commitment to legislate | No SI bill appeared in the May 13, 2026 King's Speech","penalties":"","source_urls":"https://iapp.org/news/a/king-s-speech-signals-diffuse-uk-digital-policy-agenda-but-no-ai-bill"},{"jurisdiction_id":"japan","jurisdiction":"Japan","measure_id":"jp-ai-promotion-act","short_name":"AI Promotion Act","full_name":"Act on the Promotion of Research, Development and Utilization of Artificial Intelligence-Related Technologies (Act No. 53 of 2025)","status":"in-force","introduced":"2025-02-28","signed":null,"effective":"2025-06-04","next_date":null,"next_milestone":null,"scope":"Mainly the national government; it sets no penalties for SI developers.","requires":"Promote research, development and use of SI technologies | Create the AI Strategic Headquarters under the Prime Minister | Government adopts an AI Basic Plan, first adopted December 23, 2025 | Remaining provisions, including the headquarters, took effect September 1, 2025","penalties":"None; the Act carries no statutory penalties.","source_urls":"https://digitalpolicyalert.org/event/30833-act-on-promotion-of-research-and-development-and-application-of-artificial-intelligence-related-technologies-entered-into-force-excepting-certain-provisions | https://cm.twobirds.com/en/insights/2025/japan/japan's-ai-governance-major-government-steps-since-the-ai-act | https://hlc.com/en/publications/post-election-japan-ai-policy-regulatoryoperational-updates"},{"jurisdiction_id":"south-korea","jurisdiction":"South Korea","measure_id":"kr-ai-basic-act","short_name":"AI Basic Act","full_name":"Framework Act on the Development of Artificial Intelligence and the Creation of a Foundation for Trust (Act No. 20676)","status":"in-force","introduced":null,"signed":"2025-01-21","effective":"2026-01-22","next_date":"2027-01-22","next_milestone":"End of the one year grace period on administrative fines indicated by the Ministry of Science and ICT (policy, not statute)","scope":"SI business operators serving Korea, including foreign firms; safety duties apply to systems above a training compute threshold set by decree, reported as 10^26 FLOP.","requires":"High-compute systems: identify, assess and mitigate risks across the lifecycle and report results to the Minister of Science and ICT | High-impact SI: pre-deployment self-review, risk management, explanation and user protection plans, and human oversight | Notify users in advance when a product uses generative or high-impact SI, and label generated outputs | Clearly mark realistic synthetic audio, images and video | Foreign operators above revenue or user thresholds appoint a domestic agent","penalties":"Administrative fines of up to 30 million won for failing to give notice, appoint a domestic agent or obey corrective orders.","source_urls":"https://www.law.go.kr/LSW/eng/engLsInfoR.do?lsiSeq=268543 | https://www.cooley.com/news/insight/2026/2026-01-27-south-koreas-ai-basic-act-overview-and-key-takeaways"},{"jurisdiction_id":"canada","jurisdiction":"Canada","measure_id":"ca-aida","short_name":"AIDA","full_name":"Artificial Intelligence and Data Act (Part 3 of Bill C-27)","status":"stalled","introduced":null,"signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Would have been Canada's first dedicated federal statute regulating SI systems.","requires":"Died on the order paper when Parliament was prorogued on January 6, 2025 | No standalone successor SI bill tabled as of September 2026 | The privacy part returned separately as Bill C-36 in June 2026, without SI provisions","penalties":"","source_urls":"https://www.fruggr.io/en/what-replaced-canada-aida-ai-regulation-2026/"},{"jurisdiction_id":"canada","jurisdiction":"Canada","measure_id":"ca-voluntary-code","short_name":"Voluntary Code","full_name":"Voluntary Code of Conduct on the Responsible Development and Management of Advanced Generative AI Systems","status":"voluntary","introduced":null,"signed":"2023-09-27","effective":"2023-09-27","next_date":null,"next_milestone":null,"scope":"Developers and managers of advanced generative SI systems that choose to sign; 46 signatories as of March 2025.","requires":"Maintain a risk management framework proportionate to the activities | Assess and mitigate risks before deployment | Developers of public systems use multiple lines of defense, including third party audits before release | Publish capabilities and limitations and offer a free way to detect generated content | Monitor for harmful use and keep a database of post-deployment incidents","penalties":"","source_urls":"https://www.osler.com/en/insights/updates/ised-s-voluntary-code-of-conduct-on-the-responsible-development-and-management-of-advanced-generativ/ | https://www.canada.ca/en/innovation-science-economic-development/news/2025/03/canada-moves-toward-safe-and-responsible-artificial-intelligence.html"},{"jurisdiction_id":"brazil","jurisdiction":"Brazil","measure_id":"br-pl-2338","short_name":"PL 2338/2023","full_name":"Projeto de Lei 2338/2023","status":"proposed","introduced":null,"signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Brazil's proposed general legal framework for SI; final scope is still being negotiated in the Chamber of Deputies.","requires":"Passed the Senate by voice vote on a substitute text on December 10, 2024 | Under review by a Chamber of Deputies special committee, with Aguinaldo Ribeiro as rapporteur | Vote expected after the October 2026 elections | After a Chamber vote, the bill returns to the Senate","penalties":"","source_urls":"https://bairogonzalez.com/en/in-the-world/ia/pl-2338-brazil-ai-legal-framework-2026/ | https://www.entercastconsulting.com.br/en/blog/brazil-ai-bill-pl-2338-delay-2026"},{"jurisdiction_id":"india","jurisdiction":"India","measure_id":"in-ai-governance-guidelines","short_name":"AI Governance Guidelines","full_name":"India AI Governance Guidelines","status":"voluntary","introduced":null,"signed":"2025-11-05","effective":null,"next_date":null,"next_milestone":null,"scope":"Non-binding national guidance for SI developers, deployers and government bodies.","requires":"Seven principles, including trust, people first, accountability, and safety, resilience and sustainability | AI Governance Group coordinates policy, supported by the Technology and Policy Expert Committee and the AI Safety Institute | Action plan items include India specific risk frameworks, incident reporting and liability regimes | Prefers flexible regulation and innovation sandboxes over a standalone SI law","penalties":"","source_urls":"https://www.tribuneindia.com/news/business/meity-unveils-india-ai-governance-guidelines-to-ensure-safe-inclusive-and-responsible-ai-adoption"},{"jurisdiction_id":"australia","jurisdiction":"Australia","measure_id":"au-national-ai-plan","short_name":"National AI Plan","full_name":"National AI Plan","status":"voluntary","introduced":null,"signed":"2025-12-02","effective":null,"next_date":null,"next_milestone":null,"scope":"Whole of government plan relying on existing technology neutral laws and guidance.","requires":"Did not proceed with ten proposed mandatory guardrails for high-risk SI | Update existing laws and issue guidance instead of a standalone SI act | Establish an Australian AI Safety Institute","penalties":"","source_urls":"https://piperalderman.com.au/insight/australia-now-has-a-national-ai-plan-now-what/"},{"jurisdiction_id":"australia","jurisdiction":"Australia","measure_id":"au-ai-safety-institute","short_name":"AU AISI","full_name":"Australian AI Safety Institute","status":"voluntary","introduced":null,"signed":"2025-11-25","effective":null,"next_date":null,"next_milestone":null,"scope":"Government technical hub working with existing regulators and the National AI Centre; operations were expected to start in early 2026.","requires":"Monitor, test and share information on emerging SI capabilities, risks and harms | Coordinate with existing regulators rather than replace them | Join the International Network of AI Safety Institutes","penalties":"","source_urls":"https://cm.twobirds.com/en/insights/2025/australia/australian-government-to-establish-ai-safety-institute"},{"jurisdiction_id":"australia","jurisdiction":"Australia","measure_id":"au-standards-for-ai","short_name":"Australian Standards for AI","full_name":"Australian Standards for AI","status":"proposed","introduced":"2026-07-15","signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Scope, thresholds, penalties and whether duties will be mandatory for all SI are not yet defined.","requires":"New Office of AI in the Department of the Prime Minister and Cabinet to design and legislate the standards | Mandatory requirements announced so far cover energy and water use of large SI data centers, plus copyright protections | Legislation expected in early 2027","penalties":"","source_urls":"https://compliancecouncil.com.au/insights/australia-ai-standards-july-2026-reversal-what-it-means/"},{"jurisdiction_id":"singapore","jurisdiction":"Singapore","measure_id":"sg-singapore-consensus","short_name":"Singapore Consensus","full_name":"The 2026 Singapore Consensus on Global AI Safety Research Priorities","status":"voluntary","introduced":"2025-05-08","signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Non-binding scientific consensus for researchers, funders and standards bodies, developed at Singapore's International Scientific Exchange on AI Safety.","requires":"Risk assessment, including loss of control risk and secure evaluation infrastructure | Developing trustworthy, secure and reliable systems through specification, design and verification | Control through monitoring and intervention after deployment | Societal resilience, added as a fourth area in 2026","penalties":"","source_urls":"https://arxiv.org/pdf/2506.20702 | https://arxiv.org/pdf/2608.14611"},{"jurisdiction_id":"council-of-europe","jurisdiction":"Council of Europe","measure_id":"coe-framework-convention","short_name":"CETS 225","full_name":"Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (CETS No. 225)","status":"passed","introduced":null,"signed":"2024-09-05","effective":null,"next_date":null,"next_milestone":null,"scope":"States and the EU that ratify it, for activities across the SI lifecycle that may affect human rights, democracy and the rule of law.","requires":"Parties adopt measures so SI activities respect human rights, democracy and the rule of law | Enters into force three months after five ratifications, at least three by Council of Europe member states | Signatories include the EU, UK, US, Canada, Japan, Israel, Norway, Switzerland and Ukraine | EU ratification deposited May 15, 2026; Bosnia and Herzegovina signed December 9, 2025 and North Macedonia May 8, 2026","penalties":"","source_urls":"https://www.coe.int/en/web/artificial-intelligence/the-framework-convention-on-artificial-intelligence | https://www.coe.int/en/web/conventions/may | https://fixtheaigap.com/research/policies/2026-05-24-council-of-europe-ai-convention-eu-ratification/ | https://en.wikipedia.org/wiki/Framework_Convention_on_Artificial_Intelligence"},{"jurisdiction_id":"united-nations","jurisdiction":"United Nations","measure_id":"un-scientific-panel","short_name":"UN Scientific Panel","full_name":"Independent International Scientific Panel on AI","status":"voluntary","introduced":null,"signed":"2025-08-26","effective":"2026-02-12","next_date":null,"next_milestone":null,"scope":"40 independent experts serving three year terms, established by resolution A/RES/79/325; advisory, with no binding powers.","requires":"Produce an annual scientific assessment of SI capabilities, risks and impacts | Present each report to the Global Dialogue on AI Governance | Members drawn from all UN regions, with no more than two per nationality | Preliminary report released July 1, 2026","penalties":"","source_urls":"https://www.un.org/independent-international-scientific-panel-ai/en/faq | https://en.wikipedia.org/wiki/Independent_International_Scientific_Panel_on_AI"},{"jurisdiction_id":"united-nations","jurisdiction":"United Nations","measure_id":"un-global-dialogue","short_name":"Global Dialogue","full_name":"Global Dialogue on AI Governance","status":"voluntary","introduced":null,"signed":"2025-08-26","effective":"2026-07-06","next_date":null,"next_milestone":null,"scope":"UN member states and stakeholders; a forum with no binding outcomes.","requires":"Regular intergovernmental and multistakeholder discussion of SI governance | First session held in Geneva on July 6 and 7, 2026 | Next session planned in New York in 2027 | Receives the Scientific Panel's annual report","penalties":"","source_urls":"https://www.un.org/independent-international-scientific-panel-ai/en/faq"},{"jurisdiction_id":"g7","jurisdiction":"G7","measure_id":"g7-hiroshima-code","short_name":"Hiroshima Code of Conduct","full_name":"Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems","status":"voluntary","introduced":null,"signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Organizations developing advanced SI systems, which report voluntarily through the OECD.","requires":"Eleven actions covering risk identification, assessment and mitigation across the lifecycle | Incident reporting and information sharing | Public reporting on risk management practices through the OECD Hiroshima AI Reporting Framework | Version 2.0 of the reporting framework adds coverage of agentic SI","penalties":"","source_urls":"https://oecd.ai/en/transparency/about | https://oecd.ai/en/haip-2-launch | https://www.soumu.go.jp/hiroshimaaiprocess/en/report.html"},{"jurisdiction_id":"international-summits","jurisdiction":"International summit series","measure_id":"bletchley-declaration","short_name":"Bletchley Declaration","full_name":"The Bletchley Declaration by Countries Attending the AI Safety Summit, 1-2 November 2023","status":"voluntary","introduced":null,"signed":"2023-11-01","effective":null,"next_date":null,"next_milestone":null,"scope":"28 countries and the EU; New Zealand joined on October 23, 2024.","requires":"Recognize potential serious, even catastrophic, harm from the most capable SI models | Flag cybersecurity, biotechnology and disinformation risks and possible loss of human control | Developers of the most powerful frontier systems bear a particularly strong responsibility for safety | Developers should use safety testing and evaluations and be transparent about mitigations","penalties":"","source_urls":"https://www.gov.uk/government/publications/ai-safety-summit-2023-the-bletchley-declaration/the-bletchley-declaration-by-countries-attending-the-ai-safety-summit-1-2-november-2023."},{"jurisdiction_id":"international-summits","jurisdiction":"International summit series","measure_id":"seoul-frontier-commitments","short_name":"Seoul Commitments","full_name":"Frontier AI Safety Commitments, AI Seoul Summit 2024","status":"voluntary","introduced":null,"signed":"2024-05-21","effective":null,"next_date":null,"next_milestone":null,"scope":"16 companies at launch, with Magic, MiniMax, 01.ai and NVIDIA added later for 20 in total.","requires":"Assess risks across the lifecycle, including before deployment | Set thresholds at which severe risks are deemed intolerable unless adequately mitigated | In the extreme, do not develop or deploy a model if risks cannot be kept below thresholds | Publish a safety framework focused on severe risks ahead of the next summit in France | Maintain internal accountability and public transparency on safety approaches","penalties":"","source_urls":"https://www.gov.uk/government/publications/frontier-ai-safety-commitments-ai-seoul-summit-2024/frontier-ai-safety-commitments-ai-seoul-summit-2024"},{"jurisdiction_id":"international-summits","jurisdiction":"International summit series","measure_id":"new-delhi-declaration","short_name":"New Delhi Declaration","full_name":"New Delhi Declaration on AI Impact","status":"voluntary","introduced":null,"signed":null,"effective":null,"next_date":null,"next_milestone":null,"scope":"Endorsed by 92 countries and international organizations, including the US, China and the EU, as of March 5, 2026.","requires":"Secure and Trusted AI pillar recognizes the security of SI systems and industry voluntary measures | Notes the Trusted AI Commons, a voluntary platform of tools, benchmarks and practices | Calls for policy frameworks that protect the public interest across the SI lifecycle | All commitments are voluntary and non-binding","penalties":"","source_urls":"https://www.pib.gov.in/PressReleasePage.aspx?PRID=2231208&reg=3&lang=1"},{"jurisdiction_id":"international-summits","jurisdiction":"International summit series","measure_id":"new-delhi-frontier-commitments","short_name":"New Delhi Commitments","full_name":"New Delhi Frontier AI Commitments","status":"voluntary","introduced":null,"signed":"2026-02-19","effective":null,"next_date":null,"next_milestone":null,"scope":"Global and Indian frontier SI companies; no signatory list was published in the source reviewed.","requires":"Use anonymized, aggregated usage data to inform policy on jobs, skills and workforce planning | Evaluate SI systems across languages and cultural contexts | Voluntary commitments announced by India's IT minister at the AI Impact Summit","penalties":"","source_urls":"https://ddnews.gov.in/en/india-unveils-new-delhi-frontier-ai-commitments-at-ai-impact-summit/"},{"jurisdiction_id":"us-china","jurisdiction":"United States and China (bilateral)","measure_id":"us-china-si-dialogue","short_name":"US-China SI Dialogue","full_name":"U.S.-China Super Intelligence (SI) Dialogue and AI incident communication channel","status":"voluntary","introduced":null,"signed":"2026-09-25","effective":null,"next_date":null,"next_milestone":null,"scope":"The two governments only; no binding obligations.","requires":"Set up a bilateral communication channel for SI related incidents | Hold a dialogue on SI risks and benefits, with the next exchange by November 2026 | Chip export controls are not part of the mechanism, according to the US Trade Representative","penalties":"","source_urls":"https://www.aljazeera.com/news/2026/9/26/china-us-to-open-ai-communication-channel-after-summit-white-house-says | https://www.pbs.org/newshour/world/china-and-u-s-agree-to-establish-ai-safety-channel-and-continue-trade-and-military-talks | https://www.secondtalent.com/resources/us-china-ai-deal/"},{"jurisdiction_id":"global-civil-society","jurisdiction":"Global civil society","measure_id":"fli-superintelligence-statement","short_name":"FLI Statement","full_name":"Statement on Superintelligence","status":"voluntary","introduced":null,"signed":"2025-10-22","effective":null,"next_date":null,"next_milestone":null,"scope":"Open to individuals and organizations; signatures carry no legal force.","requires":"Prohibit the development of superintelligence | Lift the prohibition only after broad scientific consensus that it can be done safely and controllably | Also require strong public buy-in before lifting it","penalties":"","source_urls":"https://superintelligence-statement.org/ | https://newsletter.futureoflife.org/p/fli-newsletter-october-2025"}],"lab_policies":[{"id":"anthropic-rsp","lab":"Anthropic","policy":"Responsible Scaling Policy","version":"3.4","updated":"2026-07-08","summary":"Version 3.0 (February 24, 2026) rewrote the policy around a public Frontier Safety Roadmap and periodic Risk Reports, specified mitigations only for the next capability level, and dropped the earlier implied pause commitment. Later versions let the Long-Term Benefit Trust request external review of Risk Reports and revised the automated R&D threshold.","url":"https://www.anthropic.com/responsible-scaling-policy"},{"id":"openai-preparedness","lab":"OpenAI","policy":"Preparedness Framework","version":"2","updated":"2025-04-15","summary":"Tracks biological and chemical, cybersecurity and SI self-improvement capabilities against High and Critical thresholds, with a Safety Advisory Group advising leadership. OpenAI announced a full rewrite on August 18, 2026 after judging a model may have reached the Critical cyber threshold; no replacement version had been located as of this update.","url":"https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf"},{"id":"openai-frontier-governance","lab":"OpenAI","policy":"Frontier Governance Framework","version":"May 2026","updated":"2026-05-28","summary":"Maps OpenAI's safety and security practices to California's frontier transparency law and the EU General-Purpose AI Code of Practice, covering cyber offense, CBRN, harmful manipulation and loss of control. It sits alongside the Preparedness Framework and does not replace it.","url":"https://openai.com/index/openai-frontier-governance-framework"},{"id":"google-deepmind-fsf","lab":"Google DeepMind","policy":"Frontier Safety Framework","version":"3.1","updated":"2026-04-17","summary":"Sets Critical Capability Levels for CBRN, cyber, harmful manipulation and ML R&D, and adds Tracked Capability Levels, including one for stealth and situational awareness tied to misalignment. Reaching a Critical Capability Level requires a safety case and governance approval before external or high-risk internal deployment.","url":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf"},{"id":"meta-superintelligence-scaling","lab":"Meta","policy":"Superintelligence Scaling Framework","version":"2.1","updated":"2026-10-02","summary":"The October 2026 revision of the framework previously titled the Frontier AI Framework sets Critical, High and Moderate risk levels across cybersecurity, chemical and biological, and loss of control outcomes. High risk models deploy only with mitigations that bring risk to moderate, and development past Critical continues only when catastrophic risks are judged acceptable.","url":"https://research.meta.ai/static/meta-superintelligence-scaling-framework"},{"id":"xai-faif","lab":"xAI","policy":"Frontier Artificial Intelligence Framework","version":"June 30, 2026","updated":"2026-06-30","summary":"Commits xAI to at least annual systemic risk assessments covering CBRN, offensive cyber, loss of control and harmful manipulation, with release only when risks are judged acceptable. It also sets security controls for model weights and an incident response process that can include shutting systems down.","url":"https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf"},{"id":"microsoft-fgf","lab":"Microsoft","policy":"Frontier Governance Framework","version":"February 2026","updated":null,"summary":"Screens Microsoft's most advanced models for CBRN, offensive cyber, advanced autonomy, loss of control and harmful manipulation capabilities, rescreening at least every six months. Risks must be brought to low or medium before deployment, or development and deployment pause.","url":"https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/msc/documents/presentations/CSR/Frontier-Governance-Framework-Feb-2026.pdf"},{"id":"amazon-fmsf","lab":"Amazon","policy":"Frontier Model Safety Framework","version":"September 2026","updated":"2026-09-17","summary":"Sets Critical Capability Thresholds for CBRN weapons proliferation, offensive cyber operations and loss of control, and commits Amazon not to deploy a model that meets a threshold until safeguards bring risk to acceptable levels. This is the reviewed update of the first 2025 version.","url":"https://cdn.amazon.science/a0/4f/bca5cb32495280835d9370a466b7/amazon-fmsf-9-2026.pdf"}]}}