The Trump White House has grown increasingly frustrated with Anthropic CEO Dario Amodei, leading to his replacement in key meetings by cofounder Tom Brown, according to people familiar with the matter. The shift comes as the administration deliberates over export controls that took Anthropic's most powerful models offline after the National Security Agency (NSA) confirmed there were ways to disable guardrails and access more powerful capabilities of the company's restricted Mythos model.
Tensions Over AI Export Controls
The administration has not yet lifted the export controls that were imposed on June 12 after the NSA's assessment. However, multiple calls have taken place in recent days, with Tom Brown and Anthropic's public policy chief Sarah Heck leading the outreach. "Tom Brown is not being a weirdo like Dario and can actually engage," said one person directly familiar with the calls, according to WIRED. Amodei, the people say, was too difficult to talk to and did not listen to concerns.
The talks have been conducted at both high-level and working-group levels, involving technical staff from both sides. Some conversations have focused on establishing what level of proof from Anthropic could alleviate the administration's concerns about jailbreaks of the Claude Fable 5 model. Independent cybersecurity experts have increasingly viewed guardrails as only a stopgap, since skilled users and future AI models will find ways to bypass constraints, as Inner Loop previously noted.
A Change in Leadership Style
The White House's preference for Tom Brown marks a significant shift in dynamics. While Dario Amodei has been a prominent voice in AI safety, his approach has apparently clashed with administration officials. The Commerce Department's Bureau of Industry and Security (BIS), which manages export controls, has taken a leading role in addressing jailbreak risks. A spokesperson for the White House declined to comment, and Anthropic did not respond to a request for comment.
| Key Fact | Detail |
|---|---|
| Export controls imposed | June 12, after NSA affirmation |
| Models affected | Claude Fable 5 (re-release), Mythos (restricted) |
| Administration preference | Tom Brown over Dario Amodei |
| Lead officials | Tom Brown, Sarah Heck (Anthropic); Howard Lutnick (Commerce) |
Path Forward Uncertain
The timeline for re-deploying Claude Fable 5 remains uncertain. Last week, a bipartisan group of lawmakers sent a list of questions to Commerce Secretary Howard Lutnick about the path forward. Among the questions was one on redeployment: "What specific criteria does the Department rely upon for determining whether to restore public access to the model through a revision of this decision? What is the timeline for that decision?"
The letter, signed by Representatives Sam Liccardo, Jay Obernolte, C. Scott Franklin, and Ted Lieu, demanded responses by June 26. A Commerce Department spokesperson declined to comment on whether the agency would respond by the deadline. For enterprise technology decision-makers evaluating AI adoption in regulated environments, this episode underscores the growing friction between AI companies and government over safety standards and export controls, which could affect the availability of advanced models for business use.
Sources:
