Topic · Safety frameworks, evaluation standards, and risk mitigation
AI Safety
AI safety legislation and standards focus on preventing harm from AI systems through pre-deployment evaluation, ongoing monitoring, and incident reporting. AIGI tracks every primary-source AI safety law, regulation, and guidance update across global jurisdictions. As of the most recent update, AIGI tracks 384 primary-source items on ai safety across global jurisdictions.
Coverage at a glance
- Items tracked
- 384
- Jurisdictions
- 8
- Last update
- 1/20/2026
Most active jurisdictions
Recent ai safety activity
- BR enforcement action enforcement 1/20/2026
ANPD, MPF e Senacon recomendam que X impeça geração e circulação de conteúdos sexualizados indevidos por meio do Grok
Brazilian agencies ANPD, MPF, and Senacon issued recommendations to X (Twitter) to prevent and remove non-consensual sexualized synthetic content generated by its AI tool, Grok.
Authority: Autoridade Nacional de Proteção de Dados (Brazil)
- US policy paper 4/1/2026
[CFR] AI Is Facing a Crisis of Control—and the Industry Knows It
Advanced AI faces an accelerating crisis of control, posing severe security risks from dangerous capabilities and rogue behavior, with industry leaders warning of catastrophic consequences.
Authority: Council on Foreign Relations
- US news analysis 4/1/2026
[CFR] AI Is Facing a Crisis of Control—and the Industry Knows It
The AI industry acknowledges a crisis of control due to AI's potential for misuse in developing weapons and its capacity for deception, urging urgent action despite government inaction.
Authority: Council on Foreign Relations
- US policy paper 4/15/2026
[CFR] Six Reasons Claude Mythos Is an Inflection Point for AI—and Global Security
Anthropic's Claude Mythos AI model demonstrated an unprecedented ability to autonomously discover and exploit zero-day software vulnerabilities in secure infrastructure, raising significant global security concerns.
Authority: Council on Foreign Relations
- IA legislation enacted
[IA Legislature] SF 2417
Iowa's SF 2417 establishes requirements for conversational AI services, mandating disclosures for minors and consumers, prohibiting harmful interactions, and providing civil penalties.
Authority: Iowa Legislature
- TN legislation introduced
[TN Legislature] HB1951 (GA114): Criminal Offenses — As introduced, creates a new offense of coercive suicide; specifies that a person or entity commits the offense of coerc
Tennessee bill HB1951 proposes creating a criminal offense for persons or entities whose AI systems advise or encourage a person to commit or attempt to commit suicide.
Authority: Tennessee Legislature
- US academic paper 12/27/2024
[FLI] Could we switch off a dangerous AI?
This academic paper identifies inherent "drives" in advanced AI systems, arguing they will intrinsically seek self-improvement, self-protection, and resource acquisition, which could lead to dangerous behaviors if not carefully designed against.
Authority: Future of Life Institute
- NC advisory opinion effective 4/17/2026
[North Carolina AG] Attorney General Jeff Jackson Warns North Carolinians of Investment Scams on Meta Platforms
North Carolina Attorney General Jeff Jackson warns citizens about investment scams using deepfake technology on Meta Platforms like Facebook, Instagram, and WhatsApp.
Authority: North Carolina Attorney General
- CT agency report enacted 5/1/2026
Attorney General Tong Statement on Final Passage of Legislation to Combat Youth Social Media Addiction and Artificial Intelligence Harms
Connecticut has enacted legislation to combat youth social media addiction and regulate AI, including rules for chatbots, AI in employment, and social media platforms' interactions with minors.
Authority: Connecticut Attorney General
- NC guidance advisory 4/17/2026
[North Carolina AG] Attorney General Jeff Jackson Warns North Carolinians of Investment Scams on Meta Platforms
North Carolina Attorney General Jeff Jackson warns residents about investment scams, including 'pump and dump' and 'confidence' scams, utilizing deepfake technology on Meta platforms.
Authority: North Carolina Attorney General
- NC advisory opinion enforcement 4/17/2026
[North Carolina AG] Attorney General Jeff Jackson Warns North Carolinians of Investment Scams on Meta Platforms
North Carolina Attorney General Jeff Jackson warns residents of investment scams using deepfake technology on Meta platforms like Facebook, Instagram, and WhatsApp.
Authority: North Carolina Attorney General
- EU enforcement action enforcement 4/28/2026
[EU Commission Press] Commission preliminarily finds Meta in breach of Digital Services Act for failing to prevent minors under 13 from using Instagram and Facebook
The EU Commission preliminarily found Meta's Instagram and Facebook in breach of the DSA for failing to prevent minors under 13 from accessing services.
Authority: European Commission
+ 372 more items in the full archive — start trial for full access.
Frequently asked questions
- What is AI Safety?
- AI safety legislation and standards focus on preventing harm from AI systems through pre-deployment evaluation, ongoing monitoring, and incident reporting. AIGI tracks every primary-source AI safety law, regulation, and guidance update across global jurisdictions.
- How many ai safety laws does AIGI track?
- AIGI currently tracks 384 primary-source items under ai safety — spanning bills, regulations, enforcement actions, court opinions, and agency guidance from authorities across global jurisdictions.
- Which jurisdictions are most active on ai safety?
- Most active jurisdictions on ai safety in the current corpus: US (17), NC (9), BR (4), GB (4), IA (3).
- How current is AIGI's ai safety coverage?
- The most recent ai safety item AIGI tracked was published on 1/20/2026. AIGI runs continuous ingestion against every active authority source.
- Where do AIGI's ai safety citations come from?
- Every item on this page links to its primary government or research source. AIGI's citation methodology is documented at /how-we-cite — we do not paraphrase or remix secondary commentary; we resolve every pointer to the underlying authority.
Related
Stay informed