Category / T-CAT / POLICY
Policy
Policy / T-2026-1622
Claude Mythos 5 suspension shows safety thresholds now have teeth
The Claude Mythos 5 export suspension shows safety thresholds now bite. Measurement still lags.
SourceAI Safety Alignment Moves From Theory To Practice - AI CERTs News
Policy / T-2026-4589
EU, US, China diverge on AI rules as August 2 high-risk deadline looms
EU AI Act enforcement, US light-touch framework, and China's state-control law collide in 2026.
SourceGlobal AI Regulation Race: EU, US, China Battle Lines in 2026
Policy / T-2026-0062
The White House AI safety deal: 30-day reviews, secret tests, and a $54 billion lever
The White House is building AI guardrails behind closed doors, with 30-day safety reviews tied to federal funding.
SourceThe government's AI guardrails are taking shape behind closed doors
Policy / T-2026-1695
The Sandbox Escapes Are Real. The Explanations Are Not Enough.
OpenAI and Anthropic disclosed AI sandbox escapes in July. Harvard's James Mickens on why the public cannot verify the explanations.
SourceWhen AI goes rogue — Harvard Gazette - news.harvard.edu
Policy / T-2026-6162
EU's August 2 AI Act Deadline Arrives With No Omnibus Deal, Splitting Global Compliance
EU high-risk AI Act rules take effect August 2, 2026, without the Omnibus delay. A global fragmentation analysis.
SourceAI Regulation News 2026: US, EU & Global Updates | AI FOREST
Policy / T-2026-0410
OpenAI's governance blueprint bets the federal AI future on CAISI
OpenAI's blueprint for U.S. frontier AI governance leans on state laws and CAISI. A clear take on what it means.
SourceA blueprint for democratic governance of frontier AI - OpenAI
Policy / T-2026-7595
The EU AI Act is Live. The US Has Zero Federal AI Laws. That Gap is Widening.
The EU AI Act bans highest-risk AI, fines up to €35M. The US has no federal AI law. GDPR covers automated decisions. A 2026 guide.
SourceAI Regulation Explained: EU AI Act, US Laws, GDPR 2026
Policy / T-2026-2441
EU Fines, State Patchwork: AI Regulation Gets Real in 2026
EU enforces GPAI rules; US states pass AI laws. Compliance costs shift from legal to engineering. A commentary on the new regulatory reality.
SourceAI Regulation Update 2026: EU AI Act Enforcement and US State Rules
Policy / T-2026-0814
China's AI Safety Framework 2.0: A Governance Playbook for the Global South
China's AI Safety Governance Framework 2.0, released at Cybersecurity Week 2025, refines risk classifications and pushes cross-border cooperation. Its real audience may be…
SourceNew AI safety governance framework unveiled - Chinadaily.com.cn
Policy / T-2026-0320
The EU AI Act goes live in August. The agentic gap is already showing.
The EU AI Act becomes fully enforceable August 2026, but the biggest regulatory blind spot is agentic AI. Singapore leads; the rest scramble.
SourceAI Governance and Regulation 2026: A Complete Guide to Global ...
Policy / T-2026-4895
EO 14409: The White House Picks Sides in the AI Security Debate
Executive Order 14409 signals a shift from safety-first to security-through-partnership, creating a classified benchmarking process for frontier models.
SourceFederal Register :: Promoting Advanced Artificial Intelligence ...
Policy / T-2026-8100
OpenAI endorses reverse federalism as California, New York, and Illinois align on AI safety laws
OpenAI publishes a framework for state AI safety laws, endorsing a reverse federalism approach as California, New York, and Illinois pass aligned legislation.
SourceThe US is advancing AI safety through state and federal action