AI Policy Newsletter - June 26, 2026

• On June 24, the Ratepayer Protection Act – legislation introduced by Reps. Gabe Evans (R-CO) and Kathy Castor (D-FL) – passed the House Energy & Commerce Energy Subcommittee by voice vote. The legislation would amend the Public Utility Regulatory Policies Act (PURPA) to require large-load customers to cover the full incremental costs of grid upgrades needed to serve their facilities, codifying the principles underlying the voluntary Ratepayer Protection Pledge commitments made by Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI to cover their own data center energy costs.

• On June 12, the Commerce Department issued an export control directive requiring Anthropic to obtain U.S. government approval for any foreign national to access its Claude Fable 5 and Mythos 5 models — a directive Anthropic says effectively forced it to disable both models for all customers globally to ensure compliance. The government cited a jailbreak vulnerability in Fable 5, while Anthropic disputed the severity of the finding and argued that applying the standard industry-wide “would essentially halt all new model deployments.” Talks between Anthropic and the administration have since shifted toward developing a common technical framework for evaluating future jailbreaks, though the models remain offline.

• On June 4, Reps. Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a 269-page bipartisan discussion draft of the Great American Artificial Intelligence Act, the most significant bipartisan legislative effort to establish a federal AI governance framework to date. The bill is organized around four titles covering frontier AI governance, workforce, cybersecurity, and research and development. Its most consequential provision is a three-year preemption of state AI laws. Despite bipartisan co-sponsorship, the bill lacks leadership backing in either chamber and has drawn opposition from labor groups and civil liberties organizations, as well as key House Democrats. Sponsors are actively soliciting stakeholder feedback with no set deadline.

• Senator Marsha Blackburn (R-TN) is working with White House AI and tech leaders to develop a legislative package addressing both artificial intelligence and protecting kids online. The bill would include limited rather than blanket preemption of state AI laws and is focused on what Blackburn calls the “four Cs”: children, content creators, communities, and censorship. The bill would also incorporate the Senate versions of the Kids Online Safety Act (KOSA) and the NO FAKES Act. Senate Commerce Chairman Cruz (R-TX) intends to hold a markup in the July work period but has not indicated precisely which proposals would be included. Given the limited number of legislative days remaining in the session and competing privacy and AI priorities in the House and Senate Commerce Committees, passage of any such proposal will be difficult.

Download the full newsletter here.