White House Weighs Plan to Bypass AI Restrictions Tied to Anthropic

• White House drafts plan to bypass Anthropic risk flag for AI onboarding.
• Pentagon tensions persist despite signals of easing ties with Anthropic.
• Lawmakers push AI rules targeting child safety and oversight of OpenAI.

The White House is weighing draft guidance that could allow federal agencies to move forward with onboarding new artificial intelligence systems despite an existing supply-chain risk designation tied to Anthropic.

The proposal, still under consideration, would create a pathway for agencies to bypass the restriction and adopt models such as Mythos, Anthropic’s latest release. The move points to a possible shift in how the administration manages its relationship with the AI developer, following months of tension linked to defense-related concerns.

Draft Guidance Targets Onboarding Flexibility

According to reports, the draft executive action could provide a mechanism to de-escalate the dispute between federal authorities and Anthropic. While details remain limited, the reported approach would enable agencies to proceed with integrating advanced AI tools without being blocked by prior risk classifications.

Anthropic declined to comment on the development, while the White House did not immediately respond to requests for clarification. However, the reported proposal comes shortly after Anthropic introduced Mythos, described as its most advanced AI system to date. Experts cited in the report noted that the system showed a strong capability to identify cybersecurity vulnerabilities and simulate exploitation pathways.

Pentagon Tensions Shape Policy Backdrop

The draft guidance follows a period of friction between Anthropic and the Pentagon. The dispute emerged after the company refused to remove safeguards that restrict the use of its AI systems for autonomous weapons or domestic surveillance. In response, the Department of Defense applied a supply-chain risk designation to the Claude developer.

Recent comments from Donald Trump suggest a possible thaw in relations. Speaking after a meeting between Anthropic CEO Dario Amodei and White House officials, Trump said the company was shaping up in the administration’s view. When asked about a potential agreement with defense authorities during an appearance on Squawk Box, he confirmed that a deal remained possible.

Despite these signals, divisions still exist within the Pentagon, where some officials remain opposed to easing restrictions. Others, according to the report, view the standoff as counterproductive and support efforts to find a resolution, though further negotiations may still be required.

Lawmakers Advance AI Oversight Measures

Separately, members of Congress from both parties introduced legislation to address AI-related risks while ensuring continued development. A proposal led by Senators Ted Cruz and Brian Schatz would require chatbot providers to offer family account features, allowing parents to monitor their children’s usage and set limits.

The legislation follows ongoing legal challenges facing OpenAI, including product liability lawsuits. In parallel, the House has advanced a separate bill requiring disclosures when companies know a user is a minor.

Another bipartisan measure introduced by Representatives Ted Lieu and Jay Obernolte focuses on expanding AI research, establishing standards, and supporting workforce cybersecurity training through tax incentives.

Related: Anthropic Takes Legal Action Against U.S. Defense Department