AI Policy · Daily

Senior U.S. officials are discussing federal equity stakes with frontier AI CEOs, with Sam Altman among those pitching the structure. Anthropic called for a verifiable industrywide pause mechanism, warning that frontier models are nearing recursive self-improvement. House lawmakers unveiled the bipartisan Great American AI Act, a discussion draft pairing a three-year state preemption with $100 million annually for a new Center for AI Standards and Innovation. The House Armed Services Committee cleared a $1.15 trillion fiscal 2027 National Defense Authorization Act after a 14-hour markup with AI-related amendments.

I.Top Stories

Senior U.S. officials hold equity stake talks with frontier AI CEOs, Altman among those pitching the structure

Senior U.S. officials are in active discussions with the chief executives of leading AI companies about taking federal equity stakes in their firms, with OpenAI CEO Sam Altman among those pitching the arrangement, NOTUS reported. Discussions have centered on having firms voluntarily cede shares to the government, with returns potentially directed to public purposes such as a dividend payment to American households. A person familiar with the matter said Anthropic is not having those conversations with the administration. The reporting follows Sen. Bernie Sanders' June 1 preview of the American AI Sovereign Wealth Fund Act, which would take a 50% public stake in the largest American AI firms through a one-time stock tax.

Read at NOTUS ↗

Anthropic calls for verifiable industrywide pause mechanism, warning frontier models near recursive self-improvement

Anthropic on Thursday published a blog post warning that frontier AI systems are approaching recursive self-improvement, the point at which AI can autonomously design and build its own successors. The company cautioned that small misalignments in today's models could compound across AI-built generations, growing more frequent but less understood until humans lose control, with development proceeding at the speed of compute rather than human oversight. Anthropic called on frontier developers to establish a verifiable, coordinated way to slow or temporarily pause development if risks rise, and said more than 80% of the code merged into its codebase as of May 2026 was generated by Claude. The Anthropic Institute will research the verification systems needed to support a credible global slowdown.

Read at Wall Street Journal ↗ • Read at Reuters ↗ • Read at Anthropic ↗

Bipartisan 'Great American AI Act' draft proposes three-year state preemption but runs into immediate opposition

Reps. Jay Obernolte (R-Calif.) and Lori Trahan (D-Mass.) unveiled a 269-page discussion draft Thursday that would preempt state laws regulating frontier AI development for three years, require developers with more than $500 million in annual revenue to publish safety frameworks and retain NIST-licensed independent auditors, and establish the Center for AI Standards and Innovation in statute with $100 million a year through fiscal 2029, Roll Call reported. Developers could face penalties of up to $1 million per day for safety or audit violations. The draft drew immediate pushback over preemption — including from the three House Democrats tapped by Minority Leader Hakeem Jeffries to lead on AI, who said it "cannot serve as the basis for productive dialogue" — and faces Republican leadership skepticism that most likely pushes any serious action to 2027, Politico reported.

Read at Roll Call ↗ • Read at Politico ↗

House Armed Services Committee advances $1.15 trillion fiscal 2027 NDAA after 14-hour markup that includes some AI guardrails

The House Armed Services Committee voted out the fiscal 2027 National Defense Authorization Act late Thursday after 14 hours of debate covering roughly 900 amendments, The Hill reported. The $1.15 trillion bill heads to the House floor with AI-related amendments covering report language on AI adoption and deployment, AI-enabled advanced manufacturing, and requiring human oversight in AI-testing and evaluation. Notable failed amendments include language addressing lethal autonomy and human-in-the-loop requirements.

Read at The Hill ↗ • Read at House Armed Services ↗

II.China Watch

Chinese team reports quantum memory breakthrough aimed at the classical data bottleneck

A team led by Zhejiang University has built what it describes as the world's first superfast quantum random access memory, an interface that lets quantum computers pull from classical datasets without losing their speed advantage, SCMP reported. The team said the device targets practical applications Beijing has flagged as priorities, including drug discovery and financial fraud detection. The result follows other recent state backed milestones such as the Jiuzhang 4.0 photonic system and the Hanyuan-2 dual-core machine. Beijing treats quantum as a strategic technology, and U.S. analysts track each Chinese milestone for dual use signals on AI and cryptography.

Read at SCMP ↗

Sanctioned Chinese satellite firm posts high resolution images of Nvidia and Apple campuses

Changguang Satellite released high resolution overhead images of Nvidia's Santa Clara headquarters and Apple Park in Cupertino on social media Monday, SCMP reported. The company operates China's Jilin-1 commercial Earth observation constellation from Changchun in northeast China. The U.S. government sanctioned the company last year for alleged assistance to Iran, and a Changguang spokesman called the release routine satellite news drawn from publicly available imagery. The Jilin-1 constellation now operates more than 100 satellites.

Read at SCMP ↗

Huawei Cloud doubles down on a fully domestic AI stack with embodied AI platform

Huawei Cloud CEO Zhou Yuefeng used the company's INSPIRE conference Friday to unveil more than a dozen new AI offerings, headlined by CloudRobo, which Huawei described as the world's first full stack embodied AI development platform, Pandaily reported. The package also included an agentic memory storage system for long-running AI agents and an upgrade to the ModelArts service that now hosts Kimi, Zhipu AI, DeepSeek and MiniMax alongside Huawei's own models. The company said CloudRobo will enter public beta this month. Huawei has been on the Commerce Department's Entity List since 2019 and now offers Chinese customers a fully domestic alternative across compute, storage, model services and embodied AI development.

Read at Pandaily ↗

Chinese university team moves laser written glass storage into small-scale mass production

Researchers at Huazhong University of Science and Technology, working with Wuhan startup YiYao Technology, said they have moved laser written glass storage discs into small-scale mass production, Pandaily reported. Each disc, encoded by femtosecond laser pulses across 400 stacked layers, holds about 360 terabytes and is designed as write-once archival media with retention measured in centuries. A former chief researcher from Microsoft's Project Silica has joined YiYao as a co-founder. YiYao is targeting magnetic tape replacement for Chinese AI data centers, which operate under both Beijing's data localization rules and continued U.S. controls on advanced memory chips.

Read at Pandaily ↗

III.Federal Policy Tracker

Defense Secretary Hegseth formally declines to lift Anthropic's supply chain risk designation, sending case to D.C. Circuit ruling

Defense Secretary Pete Hegseth denied Anthropic's request to reconsider its designation as a national security risk, the Pentagon told the D.C. Circuit Court of Appeals on Thursday, Politico reported. The move clears a three-judge panel — including two Trump appointees — to rule on Anthropic's challenge to the designation and the scope of Pentagon authority over domestic firms, a label historically reserved for foreign companies tied to U.S. adversaries. In his June 3 decision, Hegseth wrote that "pre-deployment risks" with Anthropic's products and a "loss of trust" remain sufficient to support the determination, rather than any ability to manipulate Claude in real time, a capability Anthropic says it does not have.

Read at Politico ↗

NSA uses Anthropic's Mythos model for offensive cyber operations

The National Security Agency has access to Anthropic's Mythos model for offensive cyber operations, the Financial Times reported Thursday. Two sources confirmed the NSA's access, and one said the model is being used more widely within the Defense Department, per Axios. Organizations with Mythos access typically use the model to scan their own environments for exploitable security vulnerabilities.

Read at Financial Times ↗ • Read at Axios ↗

Sen. Warren invites Nvidia CEO Jensen Huang to June 11 Senate Banking hearing on China chip sales

Senate Banking ranking member Sen. Elizabeth Warren invited Nvidia CEO Jensen Huang to testify before the committee on June 11, asking him to confirm attendance by Monday, CNBC reported. In a letter sent Thursday, Warren said the appearance would let Huang address Nvidia's views on U.S. export control laws and its China business. Previewing her concerns Wednesday on CNBC, Warren said the chips at issue are used for military purposes in China and warned that U.S. firms profiting from the sales undermine long-term security.

Read at CNBC ↗

IV.Capability & Research Watch

V.Industry & Market Watch

a16z- and OpenAI-aligned lobbyists who fought Illinois SB 315 safeguards now back them, with the shift attributed to 2028 primary pressure

a16z- and OpenAI-aligned lobbyists who opposed state AI safety audits a year ago are now backing the safeguards in Illinois SB 315, Puck reported. Puck attributed the reversal to public opinion shifts and 2028 presidential primary pressure from named Democratic hopefuls. OpenAI endorsed SB 315 last week, calling the bill a thoughtful approach, per PYMNTS. Gov. JB Pritzker has said he will sign the measure, which mandates frontier AI safety frameworks, transparency reports and annual third-party audits.

Read at Puck ↗ • Read at PYMNTS ↗

Arizona rate case tests whether data centers or households absorb AI's power buildout costs

Arizona Public Service has filed a rate case proposing a 45% rate increase for large load customers including data centers and a 14.5% increase for households, the Wall Street Journal reported. The case is before the Arizona Corporation Commission, with administrative law judge recommendation expected in November and a commission vote in December, per the Arizona Capitol Times. APS' proposal also raises grid access charges for rooftop solar customers and seeks the company's first formula rate structure.

Read at Wall Street Journal ↗ • Read at Arizona Capitol Times ↗

Brookings analysis maps highest AI exposure U.S. counties to Democratic leaning vote, framing 2026 midterm AI politics for the left

A Brookings analysis mapped AI worker exposure by U.S. county and found the highest exposure counties voted Democratic in 2024, Fortune reported. Brookings finds the “techlash” against artificial intelligence is spreading, driven by workers’ fears of the technology’s potential to disrupt their jobs and upend their livelihoods.

Read at Fortune ↗ • Read at Brookings ↗

Nvidia certifies Samsung, SK Hynix and Micron to supply HBM4 memory for Vera Rubin platform

Nvidia CEO Jensen Huang confirmed Samsung, SK Hynix and Micron have all been certified to supply HBM4 high bandwidth memory for the company's Vera Rubin AI accelerators, Bloomberg reported. Huang made the confirmation in Seoul during a trip to meet Korean chaebol leaders. Industry analysts had estimated SK Hynix would hold the largest HBM4 allocation at 60% to 70%, with Samsung at 25% to 30% and Micron supplying the remainder, per TechTimes.

Read at Bloomberg ↗ • Read at TechTimes ↗

VI.Global & Geopolitics

Canadian PM Carney releases national AI strategy with C$500 million ($360 million) Tech Growth Fund and equity stake authority

Canadian Prime Minister Mark Carney released the country's national AI strategy Thursday, the New York Times reported. Carney warned that foreign AI platforms could be weaponized against Canadians, pointing to U.S. tech giants. The strategy includes a C$500 million ($360 million) Tech Growth Fund to help homegrown AI firms, and allows the federal government to take equity stakes in Canadian AI companies, per Yahoo Finance UK. Carney said the strategy aims to add 250,000 jobs by 2031 and lift Canadian GDP by 3%.

Read at New York Times ↗ • Read at Yahoo Finance UK ↗ • Read at Prime Minister of Canada ↗