Anthropic Sends Safety Team to White House to Fight AI Model Ban

Anthropic, one of the leading artificial intelligence companies, has sent a team of senior safety researchers to Washington to negotiate directly with White House officials. The move comes after the Trump administration issued an executive action last Friday banning all foreign access to the company's two most powerful AI models, Fable 5 and Mythos 5, citing national security concerns.

Safety Researchers Turn Diplomats

The high-stakes delegation includes Nicholas Carlini, a prominent security researcher known for his work on AI vulnerabilities, Logan Graham, who leads the team evaluating Anthropic's models for potential risks, and Dave Orr, the company's head of safeguards. Their goal is to demonstrate that Anthropic's safety measures are robust enough to justify restoring public access to the suspended models.

The export controls were triggered after scientists at Amazon, one of Anthropic's major investors, discovered that Fable 5's built-in protections could be partially bypassed. Large language models, often called LLMs, are AI systems trained on massive datasets to understand and generate human-like text. Safeguards are the safety filters that developers build into these models to prevent harmful or dangerous outputs. Anthropic disputes the severity of the findings, arguing that similar capabilities already exist in other widely deployed AI models from competitors like OpenAI.

What Is at Stake for Anthropic

The scope of the ban is unusually broad, affecting all foreign governments, companies, and individuals. Because the restrictions also covered foreign-born employees working inside the United States, Anthropic determined that a complete suspension of both models for all users was the only practical option.

The timing could not be worse. Anthropic recently filed a confidential IPO prospectus revealing a forty-seven billion dollar annual revenue run rate and a valuation approaching nine hundred sixty-five billion dollars. A prolonged shutdown of its flagship models could shake investor confidence at a critical moment.

Anthropic co-founder Tom Brown and public policy chief Sarah Heck have joined weekend calls with Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross, but the two sides have not yet reached an agreement. Both models remain offline for all users worldwide as negotiations continue.

The outcome of these talks could set a major precedent for how governments regulate the most advanced AI systems and whether technical safety teams can shape policy at the highest levels of government.