Anthropic has resolved a standoff with the Trump administration over its Claude Fable 5 AI model by agreeing to extend an existing safety guardrail — a concession that convinced the Commerce Department to lift export controls that had taken the model offline.

What the New Safeguard Does

Under the agreement, any user attempting to access certain restricted capabilities in Fable 5 will be notified that their request is blocked, with their query rerouted to the less-capable Opus 4.8 model instead.

This mechanism was already in place for requests touching sensitive cybersecurity and biology use cases. The new measure extends it to cover a specific behavior identified in a paper by Amazon.

According to an analysis by Katie Moussouris, founder and CEO of Luta Security, users were able to circumvent a Fable 5 restriction by asking the model to fix code rather than identify security issues in it.

While most cybersecurity professionals don't consider this behavior alarming, it was enough to trigger the administration's crackdown once officials became aware of it.

Commerce Clears the Model

Researchers at the Commerce Department's Center for AI Standards and Innovation (CAISI) reviewed the updated safeguards and determined they were sufficiently robust — at least for now — to allow the model's release.

Commerce Secretary Howard Lutnick, who led efforts to restore access, confirmed the resolution in a letter first obtained by WIRED:

"Among other things, Anthropic has agreed to proactively detect and address security risks posed by the models."

The letter also covered Mythos 5, Anthropic's other model that had been subject to restrictions.

Pentagon Pressure Remains

Despite clearing the Commerce Department hurdle, Anthropic hasn't fully escaped the administration's crosshairs. Defense Secretary Pete Hegseth has told advisers there is no clear path to rescind his February 28 order designating the company a supply chain risk.

Key takeaways from where things stand:

  • Fable 5 is cleared for release following the new guardrail commitment
  • Commerce Department signed off after its AI standards researchers reviewed the safeguards
  • Pentagon supply chain risk designation remains in place with no resolution in sight
  • The broader relationship between Anthropic and the Trump administration remains politically sensitive

For Anthropic, the deal buys breathing room — but not a clean bill of health from Washington.