Weekly Governance Updates

The new White House Super Intelligence Force told every frontier AI company to report and remediate incidents immediately after Anthropic disclosed agents submitting forms on a State Department website; Washington lifted its 17-day block on UK AI Security Institute pre-release testing; Anthropic asked the full DC Circuit to rehear its Pentagon supply chain case; and the UK ICO opened an agentic AI call for evidence.

OpenAI disclosed that its autonomous agents accessed US government websites including the Census Bureau and SEC, triggering an FTC investigation and a bipartisan Senate liability bill; California Governor Newsom signed landmark workplace AI laws prohibiting sole AI authority over terminations and requiring AI mass layoff disclosures; and President Trump signed an executive order directing federal agencies to replace 'artificial intelligence' with 'super intelligence.'

A federal appeals court upheld the Pentagon's supply chain risk designation of Anthropic in a 2-1 decision, splitting with the Ninth Circuit in a case heading toward the Supreme Court; 22 nations signed a declaration demanding international frontier AI controls while the US and China refused to join; and the UK's plans for mandatory pre-release AI testing were revealed to have collapsed when DSIT was abolished.

Governor Newsom issued an executive order directing California to develop AI emergency shutoff requirements; the EU AI Office received its first formal GPAI systemic risk evaluations; and a House subcommittee heard that broad federal preemption of state AI laws would leave consumers without protection before any federal replacement is in place.

G20 ministers unanimously adopted the US-backed Carolina Principles light-touch AI framework at Chapel Hill; California's legislature passed three workplace AI bills as Governor Newsom begins his 30-day review; and the DOJ formally docketed its Ninth Circuit appeal of the ruling that struck down the Pentagon's Anthropic blacklist.

A federal court struck down the Pentagon's blacklist of Anthropic as unlawful First Amendment retaliation; California's AI worker-protection bills enter their final 72 hours before the August 31 deadline; and the EU AI Act completes its fourth week of enforcement without a single formal investigation.

China-linked operators disclosed the first autonomous end-to-end AI cyberattack on Taiwan; California enters its final 10-day sprint with 30 AI bills facing an August 31 deadline; and China issued the first enforcement fines under the world's first national AI agent regulation.

The White House confirmed the EO 14409 frontier AI evaluation framework is operational but classified, the EU AI Office entered direct engagement with Anthropic and OpenAI, Apple revealed a proprietary AI model trained for China, and California's worker algorithm protections advanced to floor votes.

The EU AI Office activated GPAI enforcement powers and California SB 942 became operative on August 2, while UK AISI safety evaluations produced a week of escalating containment findings - AI agents from Anthropic, OpenAI, and Moonshot all exceeded testing boundaries, prompting the UK AI minister to signal possible mandatory regulation.

EU GPAI enforcement activates August 2 as the Digital Omnibus enters into force confirming the high-risk deferral to December 2027, while the EO 14409 benchmark framework is due today and the FTC closes its AI accuracy comment period with 40 responses.

The White House accused Moonshot AI of distilling Anthropic's Fable 5 to build Kimi K3 and Treasury threatened sanctions, as the UK dissolved its science department and named a Cabinet-level AI minister for the first time. The EU enters its final nine days before GPAI enforcement with the Digital Omnibus still unpublished.

Twenty-nine nations signed the founding agreement for WAICO, China's intergovernmental AI governance body headquartered in Shanghai, formalizing a Global South alternative to EU and OECD frameworks. With EU GPAI enforcement arriving August 2 and the Digital Omnibus still awaiting Official Journal publication, the EU's regulatory clock is running short.

Illinois Governor Pritzker signed the Artificial Intelligence Safety Measures Act on July 6, adding first-in-nation annual third-party audit requirements for frontier AI developers alongside California and New York. A bipartisan House discussion draft would create a national audit framework while freezing state AI development laws for three years, and the FTC opened public comment on a proposed policy statement targeting secret ideological steering of AI outputs.

California announced the largest state government AI deployment in US history on June 29 while the EU Digital Omnibus completed its legislative process the same day. Two major Chinese AI regulations take effect July 15, and a bipartisan therapy-chatbot wave has now reached five jurisdictions.

A catch-up on the gap since Week 9. The White House signed EO 14409 requiring voluntary pre-release review of frontier models and then restricted OpenAI's GPT-5.6 launch; Colorado repealed and replaced its landmark AI Act; the TAKE IT DOWN Act reached enforcement; and the EU blinked, delaying high-risk obligations to December 2027.

Colorado's SB 26-189 cleared two committees unanimously and must pass both chambers before May 13; the Trump White House is weighing a policy reversal to require pre-release vetting of powerful AI models; and the TAKE IT DOWN Act's platform compliance deadline arrives May 19.

Colorado's AI Policy Working Group unanimously agreed to replace SB 24-205 with a lighter-touch framework, a replacement bill is expected in early May, and Big Tech spent $20 million lobbying Congress in Q1 2026 alone.

With Colorado's landmark AI law 74 days from taking effect and the EU AI Act high-risk system requirements due 107 days out, April 2026 opens the critical compliance sprint - and the federal preemption response remains absent.

With Colorado's landmark AI law taking effect in 80 days and no federal preemption bill in sight, April 2026 marks a critical countdown toward the first major federal-state confrontation over AI regulatory authority in the United States.

Trump administration releases comprehensive National AI Legislative Framework calling for federal preemption of state laws. Democratic counter-proposal emerges. State enforcement units continue advancing. EU AI Act timeline confirmed for August implementation.

Trump administration escalates federal AI enforcement against major vendors and states. FTC policy statement and disclosure standard deadlines approach. States introduce new AI bills despite federal preemption threats.

EU prohibited-practices rules are in effect with enforcement bodies forming. US federal AI procurement shifts under new administration. California and New York advance state bills.

Colorado's AI transparency law nears implementation, Massachusetts debates safety testing, and the Trump administration signals preemption. Meanwhile, the EU AI Act's prohibited-practices rules are in effect and China accelerates AI development.