Inside the fight over Claude Mythos 5
Anthropic is locked in a dispute with the Trump administration over its Mythos 5 model, which was globally disabled after a U.S. export-control order. Meetings are underway to restore access, but tensions remain over compliance and cybersecurity risks. Industry experts warn the move could weaken American AI dominance and harm defensive cyber tools.
What changed
New talks between Anthropic and the Trump administration signal a push for a negotiated resolution, though core disagreements persist over export controls and model restrictions.
Live updates
-
Anthropic-Mythos 5 standoff: Trump admin seeks truce as export ban sparks industry alarm
confidence 93%Anthropic is locked in a dispute with the Trump administration over its Mythos 5 model, which was globally disabled after a U.S. export-control order. Meetings are underway to restore access, but tensions remain over compliance and cybersecurity risks. Industry experts warn the move could weaken American AI dominance and harm defensive cyber tools.
What's confirmed:
- Anthropic received a U.S. export control directive on Friday to suspend all access to Fable 5 and Mythos 5 by any foreign national, triggering a global shutdown.
- The Trump administration is meeting with Anthropic to discuss restoring access to its powerful models, with officials seeking a truce after the abrupt disablement.
- Cybersecurity experts warn that limiting Fable 5—designed for advanced defensive work—could inadvertently aid attackers while weakening U.S. cyber defenses.
- Anthropic’s compliance with the export order came after a reported jailbreak incident involving Mythos 5, though details remain undisclosed.
- The dispute has reignited concerns about U.S. AI regulation, with critics arguing the export ban could undermine American leadership in AI development.
- Anthropic disabled both Fable 5 and Mythos 5 worldwide in response to the Commerce Department’s order, affecting all customers regardless of nationality.
- Personality clashes within Anthropic’s team reportedly contributed to internal instability before the models were taken offline.
Still unconfirmed:
- Anthropic’s CEO has privately suggested the incident highlights the need for 'sovereign AI' models, though no formal policy shift has been announced.
- Internal documents allegedly reveal tensions between Anthropic and U.S. officials over export enforcement, but no verified leaks have been confirmed.
- A 90-minute ultimatum was issued by Amazon to Anthropic before the shutdown, though the context and specifics remain unclear.
- SpaceX is reportedly in advanced talks with Anthropic for a $40 billion compute alliance, though no official deal has been finalized.