OpenAI splits with WH on AI safety | AI CEOs ask bioweapons law | AI policy coalition writes to Armed Services leaders

Bottom Line Up Front

OpenAI urged Congress to mandate civilian safety reviews of frontier models, breaking with the voluntary review structure in Trump's executive order. The CEOs of OpenAI, Anthropic, Google DeepMind and Microsoft signed a joint letter asking lawmakers to require gene synthesis providers to screen customer orders for biological weapon sequences. A coalition of AI policy groups wrote to Armed Services leaders seeking tighter lethal autonomous weapons rules in the fiscal 2027 National Defense Authorization Act as House markup begins Thursday.

I.Top Stories

OpenAI's policy paper asks Congress to mandate civilian-led safety reviews of frontier models, breaking from Trump's voluntary EO

OpenAI released a policy paper titled "Democratic Governance of Frontier AI: A blueprint for a federal framework" that calls on Congress to require mandatory pre-release evaluations of advanced AI models and to put oversight inside civilian agencies, Politico reported. The framework splits from the executive order President Trump signed Tuesday, which created only a voluntary 30-day government review and assigned an enhanced role to the intelligence community. A company statement said Altman, who traveled to Washington for the rollout, would argue against any U.S. government approval requirement before model release.

Read at Politico ↗ • Read at Semafor ↗ • Read at OpenAI ↗

OpenAI, Anthropic, Google DeepMind and Microsoft CEOs co-sign letter urging Congress to require gene synthesis screening

Sam Altman, Dario Amodei, Demis Hassabis and Mustafa Suleyman jointly signed a public letter urging Congress to require DNA and RNA synthesis companies to screen customers and orders against sequences that could be used to design biological weapons, the Wall Street Journal reported. The letter was organized by the Institute for Progress and the Foundation for American Innovation and says there is "a real possibility" AI will lower the barriers to obtaining biological weapons, per Resultsense. Stanford biosecurity expert David Relman, a signatory, warned that AI tools can help users alter an order so that screeners cannot detect it. A bipartisan Senate bill would mandate such screening, though several signatories said labs should also screen their own users.

Read at WSJ ↗ • Read at Resultsense ↗

AI policy coalition presses Armed Services chairs for lethal autonomous weapons guardrails in NDAA as House markup opens

Americans for Responsible Innovation, the Alliance for Secure AI and the AI Policy Network jointly wrote to the House and Senate Armed Services Committee leaders Wednesday calling for stronger guardrails on lethal autonomous weapons in the fiscal 2027 National Defense Authorization Act, The Hill reported. ARI Executive Director Eric Gastfriend said the current draft takes initial steps on AI workforce, model deployment and autonomous weapons policy but does not push the Pentagon hard enough on AI talent, testing and human oversight standards. Sen. Elissa Slotkin told NOTUS she expects her March bill barring AI driven nuclear launch and domestic AI surveillance to be folded into the base text of the Senate NDAA, with Senate Armed Services markup set for next week. The House Armed Services Committee marks up its NDAA version Thursday.

Read at The Hill ↗ • Read at NOTUS ↗ • Read at ARI ↗

AI industry shifts lobbying push from White House to Capitol Hill with state law preemption as the prize

Frontier AI companies intensified their Capitol Hill lobbying campaign in the 24 hours after Trump's executive order, pivoting from the White House to Congress with federal preemption of state AI laws as the priority, Semafor reported. OpenAI is pushing a state-by-state strategy under which key states would pass uniform AI legislation, building a de facto national framework while federal action stalls. An OpenAI spokesperson confirmed Altman will attend this month's G7 summit in France.

Read at Semafor ↗

Rep. Sara Jacobs introduces Sectoral AI Governance Act giving agencies a framework to write AI rules under existing laws

Rep. Sara Jacobs (D-Calif.) introduced the Sectoral AI Governance Act, legislation that would give federal agencies a consistent framework for issuing AI rules whenever an algorithmic system is likely to contribute to violations of existing federal laws, Politico reported. The bill would let agencies like the Department of Housing and Urban Development require testing of rental screening tools for discriminatory patterns under the Fair Housing Act, according to Jacobs' office. ARI President Brad Carson and Andrew W. Marshall Fellow Owen Daniels of CSET endorsed the bill at introduction.

Read at Politico ↗ • Read at Sara Jacobs ↗

Monterey Park voters approve first U.S. ballot-initiative ban on data centers

Residents of Monterey Park, California, voted Tuesday to permanently ban data centers within the city, the first such ban approved via ballot initiative in the U.S., The Guardian reported. Early results show a decisive victory for the prohibition. Many U.S. cities and counties have passed temporary or indefinite data center moratoriums through local council action, but Monterey Park is the first to do so by direct vote.

Read at The Guardian ↗

II.China Watch

DeepSeek nears first outside funding round with Tencent and CATL as lead investors

Chinese AI lab DeepSeek is poised to raise about 50 billion yuan ($7 billion) in its first external funding round, with Tencent and battery giant CATL leading, TechNode reported. The financing would value the firm at 350 billion to 400 billion yuan ($49 billion to $56 billion) post-money, the report said. Tencent plans to commit roughly 10 billion yuan ($1.5 billion) and CATL about 5 billion yuan ($740 million), while founder Liang Wenfeng is expected to put in 20 billion yuan ($3 billion) of his own capital. China's National AI Industry Investment Fund, NetEase and JD.com are in final-stage talks to join the round. State fund participation alongside corporate backers pulls DeepSeek deeper into Beijing's national AI investment push and formalizes a national champion posture already under U.S. Commerce Department and Entity List scrutiny.

Read at TechNode ↗

Alibaba opens Qwen assistant to outside brands, with KFC, Luckin and China Eastern as launch partners

Alibaba-backed Qwen is opening its consumer app to third-party agents and skills, letting brands operate inside the assistant rather than just appear as referenced results, Pandaily reported. Early partners include Luckin Coffee, KFC, Mixue and China Eastern Airlines, with users able to order food and arrange travel by natural language request. Future agents will retain user preferences and surface proactive reminders for trips, memberships and repurchases, the report said. The move repositions Qwen from an answer engine to an action layer over Chinese consumer services. As agentic transactions scale, this layer will fall under the Cyberspace Administration of China's generative AI and algorithm registration regimes and the State Administration for Market Regulation's consumer protection oversight.

Read at Pandaily ↗

NAURA unveils a domestic atomic-level etcher, claiming to break a U.S.-Japan hold on a critical chip equipment niche

Chinese semiconductor equipment maker NAURA Technology Group unveiled a 12-inch gas cluster ion beam etching system, claiming to break a longstanding U.S.-Japan grip on atomic-scale precision etching, Pandaily reported. The technology uses clusters of gas atoms rather than energetic single ions, enabling low-damage material removal needed for advanced semiconductor nodes, MEMS devices and optoelectronics, the report said. The 12-inch designation refers to compatibility with 300mm wafers, the standard format for leading edge fabs. Pandaily said NAURA has not disclosed specific technical specifications or customer qualifications for the tool. The unveiling marks another step in Beijing's drive for self-sufficiency in fabrication gear as U.S. and allied export controls continue to constrain China's access to leading edge lithography and process tools.

Read at Pandaily ↗

BYD confirms humanoid robot program and floats its dealer network as a future sales channel

Chinese EV giant BYD is developing humanoid robots, TechNode reported, citing Yicai and confirmation from Executive Vice President Li Ke. Li said automotive AI and robotics share common technical foundations and that BYD could pursue an open robotics platform while partnering with other robotics firms. She suggested BYD could use its dealership network as a sales channel if humanoid robots eventually become consumer products. Another vertically integrated Chinese manufacturer entering embodied AI accelerates Beijing's humanoid robot industrial push and intensifies the dual use governance debate facing U.S. and allied policymakers on robotics platforms.

Read at TechNode ↗

III.Federal Policy Tracker

House Energy and Commerce subcommittee splits party-line on SECURE Data Act federal preemption bill

A House Energy and Commerce subcommittee divided along party lines Wednesday at a hearing on the SECURE Data Act, which would preempt more than 20 state comprehensive privacy laws including California's Delete Act, Illinois's Biometric Information Privacy Act and Washington's My Health, My Data Act, StateScoop reported. Bill authors Reps. John Joyce (R-Pa.) and Brett Guthrie (R-Ky.), who chairs the full Energy and Commerce Committee, framed the legislation as a pro-innovation national framework built on more than 250 written submissions and consultations with about 170 organizations. Democrats and civil-liberties advocates argued the proposed standard would be "weaker than the weakest state law."

Read at StateScoop ↗

Treasury Secretary Bessent defends voluntary AI EO cyber clearinghouse at Senate Finance

Treasury Secretary Scott Bessent defended the new AI executive order's voluntary cyber cooperation framework during testimony before the Senate Finance Committee Wednesday, FedScoop reported. Bessent told Sen. Mark Warner the order "strikes a very good balance between innovation and safety" and disputed framing that it had been scaled back from earlier drafts. The clearinghouse Bessent will stand up within 30 days is designed to coordinate vulnerability scanning across AI firms and critical infrastructure operators.

Read at FedScoop ↗

Colorado Gov. Polis vetoes state surveillance pricing ban that would have covered wages and consumer goods

Colorado Gov. Jared Polis vetoed a bill Tuesday that would have banned companies from using surveillance pricing to set workers' wages and consumer goods prices, The Guardian reported. The measure would have gone beyond Maryland's April law, which limits its ban to grocery stores, by extending to algorithmic worker-wage and broader consumer-goods pricing. Consumer advocates accused Polis of siding with dominant corporations over workers.

Read at The Guardian ↗

IV.Industry & Market Watch

SpaceX wins Grimes County, Texas tax break for $55B Terafab chip plant despite resident opposition

Grimes County, Texas commissioners voted Wednesday to grant tax incentives for SpaceX's proposed Terafab chip manufacturing project, despite organized resident opposition and threats of litigation, the Financial Times reported. SpaceX and partner Tesla will invest $55 billion initially, rising to $119 billion if fully built out, per Reuters' coverage on Yahoo Finance. The vote followed a hearing that drew more than 100 residents to the Grimes County courthouse, with speakers warning the project could strain water and power supplies and alter the rural county's character. The reinvestment zone makes SpaceX eligible for tax abatements at the proposed site near the Gibbons Creek Reservoir.

Read at FT ↗ • Read at Yahoo Finance ↗

U.S. trade groups press Treasury and Commerce on AI driven memory chip shortage

A coalition of U.S. trade groups including the Alliance for Automotive Innovation, the National Retail Federation, the Medical Device Manufacturers Association, NCTA, and the Telecommunications Industry Association wrote to the Treasury and Commerce departments Wednesday warning of "an urgent imbalance in the market for memory chips" tied to AI data center demand, Bloomberg reported. The letter said the surge in AI memory consumption is driving price increases for everyday consumer electronics, telecom infrastructure, and medical devices. The signatories urged the Trump administration to act to expand supply, building on existing federal subsidies for domestic producers including Micron.

Read at Bloomberg ↗ • Read at Investing.com ↗

CNBC investigation flags Chinese-made circuit boards beneath U.S. AI chips as a national security gap

Nearly all printed circuit boards used with Nvidia and other AI chips are made in China, CNBC reported, raising security concerns as the U.S. PCB production share has dropped from 30% to 4% while China now makes 60%. The Department of Defense is requiring most of its purchases to come from a shrinking base of domestic factories, with Assistant Secretary of War for Industrial Base Policy Mike Cadenazzi naming chips, substrates and PCBs as multiple avenues of attack. TTM Technologies shares are up nearly 500% over the past year and Sanmina has tripled, driven by AI and military circuit board demand. U.S. lawmakers are weighing new subsidies and domestic sourcing requirements.

Read at CNBC ↗

Amazon engineers tell Seattle City Council to regulate AI data center buildout

Three Amazon software engineers told the Seattle City Council on Wednesday that local governments should set the terms for data center buildout, in what activists describe as the first time Big Tech employees have publicly called for regulations governing the projects, WIRED reported. AWS software engineer Patrick Schloesser told officials the company is spending about $200 billion on capital this year, most of it on data centers and AI, while it has cut 30,000 corporate jobs since October, per CNBC. The council's Land Use and Sustainability Committee advanced a one-year moratorium on new large-scale data centers.

Read at WIRED ↗ • Read at CNBC ↗

V.Global & Geopolitics

European Commission adopts Technological Sovereignty Package with Chips Act 2.0, Cloud and AI Development Act, and Open Source Strategy

The European Commission adopted the European Technological Sovereignty Package on Wednesday, a four-part proposal combining the Chips Act 2.0, the Cloud and AI Development Act (CADA), the EU Open Source Strategy and a Strategic Roadmap for Digitalisation and AI in Energy, the New York Times reported. The package aims to double the EU's global semiconductor market share to 20% by 2030 and triple European data center capacity over five to seven years, per The Washington Post. Critical public contracts will require vendors to use EU made software and hardware, Commission Executive Vice President Henna Virkkunen said, citing "kill switch" risks from foreign providers after Microsoft suspended the email account of the International Criminal Court's top prosecutor following Trump administration sanctions. The proposal now goes to the European Parliament and the Council of the EU.

Read at NYT ↗ • Read at Washington Post ↗

UK MP Jess Asato sues Musk's xAI over deepfake sexual images in UK test case on AI system liability

British Labour MP Jess Asato filed a High Court claim in England against Elon Musk's xAI Wednesday, alleging the Grok platform produced fake sexualized images and a video of her, the Financial Times reported. Asato, a member of Prime Minister Keir Starmer's party, is seeking damages, a formal acknowledgement of illegality, and an injunction requiring xAI to stop further illegality. Law firm AWO said the case is one of the first to test legal liability for the design of an AI system itself rather than for user misuse. Grok is already subject to regulatory probes in several jurisdictions, and the City of Baltimore filed a similar suit in March.

Read at FT ↗ • Read at Investing.com ↗

India's Supreme Court publishes draft AI-in-courts rules banning AI from verdicts, bail and sentencing decisions

India's Supreme Court released draft "Regulations for Use of Artificial Intelligence in Courts, 2026" Wednesday, opening a public consultation through June 20, the Indian Express reported. The draft permits AI for legal research, citation verification, drafting assistance, translation and court administration but bars the technology from influencing decisions of law and fact, bail eligibility and sentencing, per LiveLaw. The framework would apply to the Supreme Court, High Courts, subordinate courts, tribunals and statutory adjudicatory bodies nationwide. The court's AI Committee built the regulations around principles of human primacy, transparency, accountability, data protection and judicial independence.

Read at Indian Express ↗ • Read at LiveLaw ↗

UK NHS data adviser urges clarity on Palantir's identifiable patient data access

Nicola Byrne, the UK government's National Data Guardian, called on NHS England to clarify the scope of Palantir's access to identifiable patient data, the Financial Times reported. The intervention follows FT reporting that NHS England granted U.S. contractor staff, including Palantir engineers, what an internal briefing described as "unlimited access" to identifiable patient data on the Federated Data Platform before pseudonymization. Palantir holds a £330 million contract to build the platform, and the deal has drawn warnings from MPs and patient data campaigners.

Read at FT ↗