Anthropic shuts down Fable, Mythos models following Trump admin directive
Ars Technica • • business
Key Points:
- Anthropic abruptly disabled access to its Mythos 5 and Fable 5 AI models following a US Commerce Department directive imposing export controls restricting their use outside the United States.
- The US government cited concerns over a potential jailbreak that bypasses safeguards related to cybersecurity, chemistry, and biology prompts, prompting a pause to strengthen national security defenses.
- Anthropic stated it has only received verbal evidence of a limited jailbreak involving minor software vulnerabilities and disagreed with the government's decision to recall the models, warning such standards could halt new AI deployments industry-wide.
- Other Anthropic models remain accessible, and the company apologized for the disruption, promising to provide further details within 24 hours.
- This development follows a recent executive order signed by President Trump encouraging voluntary government security testing of AI models amid ongoing regulatory discussions.