AI Policy Newsletter - March 6, 2026

• On March 5, Bloomberg reported that Department of Commerce officials have drafted regulations that would require companies to seek the administration’s permission for virtually all exports of AI chips. According to Bloomberg, the specific approval process would depend on how much computing power a company wants, with companies needing pre-clearance before seeking export licenses and potentially facing conditions such as disclosing their business models or allowing the US government site visits. For massive deployments, the US would only approve exports to allies that make stringent security promises and “matching” investments in American AI. The draft rules come as the Trump administration seeks an alternative for the Biden-era AI Diffusion Rule while working to promote the American AI stack abroad. Axios later reported that the draft rules were signed off on by Secretary of Commerce Howard Lutnick and sent to the Office of Management and Budget (OMB).

  
  

• On March 4, President Trump convened major tech leaders to formalize the Ratepayer Protection Pledge, a voluntary commitment requiring signatories (Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI) to fully fund electricity infrastructure for their AI data centers. The agreement addresses mounting concerns about AI's energy demands by ensuring companies build, bring, or buy their own power generation and cover all grid upgrades, preventing utility bill increases for American households. Companies will pay negotiated rates whether they use the electricity or not, make backup generation available during emergencies, and commit to local workforce training.

  
  

• On March 5, Secretary of War Pete Hegseth designated Anthropic a "supply chain risk" – a label typically reserved for adversarial foreign entities – effectively blacklisting the firm from federal contracting and potentially forcing partners like Nvidia to sever commercial ties. This designation follows through on President Trump’s directive last week that all federal agencies cease using Anthropic’s technology within six months after the company refused to remove ethical guardrails prohibiting the use of its AI tool, Claude, for mass domestic surveillance and fully autonomous weapons.

  
  

  • OpenAI subsequently announced a new deal to deploy its models on the Pentagon’s classified networks, claiming its agreement includes similar – and even more stringent – prohibitions on using its technology to conduct mass surveillance and autonomous force.

    
    

  • The Pentagon-Anthropic battle drew sharp bipartisan criticism from top Senate defense leaders – Senators Roger Wicker (R-MS), Jack Reed (D-RI), Mitch McConnell (R-KY), and Chris Coons (D-DE). In a letter to Secretary Hegseth and CEO Amodei, they warned that labeling Anthropic a supply chain risk “without credible evidence” could benefit China and asserted that "contract negotiations are not the ideal context in which to establish policy."

Download the full newsletter here.