Key facts
- Anthropic CEO Dario Amodei has clashed with the White House over AI policy and messaging.
- The White House imposed export controls on Anthropic's latest AI model, Fable 5.
- Anthropic removed Fable 5 from access following the export controls.
- The dispute stemmed from a reported security breach and a disagreement over the model's capabilities and guardrails.
- Both Anthropic and administration officials are engaged in negotiations to resolve the situation.
Anthropic CEO Dario Amodei has found himself in a political battle with the White House over the company's artificial intelligence models, diverging from typical tech industry deference to administration messaging. This has led to a significant rift, culminating in the White House imposing export controls on Anthropic's latest and most powerful AI model, Fable 5.
The conflict escalated after Amodei repeatedly voiced concerns about AI's potential for mass job losses and clashed with Defense Secretary Pete Hegseth over its battlefield applications. The administration also disputed Anthropic's account of a security breach on Fable 5, which was described as a "jailbreak" that bypassed the model's safety guardrails. This led to Treasury Secretary Scott Bessent, National Cyber Director Sean Cairncross, and Commerce Secretary Howard Lutnick imposing export restrictions, which Anthropic countered by taking the model offline entirely.
Despite these clashes, Anthropic has previously engaged with the White House, attending events and supporting the administration's AI Action Plan. However, the company views the recent export controls as a disproportionate response to the security risk. Negotiations are underway to resolve the issue, with a potential resolution involving a technical assessment of the vulnerability. Some critics, like former Trump official Dean Ball, suggest Amodei needs to adopt a more conciliatory approach to navigate Washington's political landscape.