The Trump administration has lifted export controls on Anthropic’s Claude Fable 5 AI model following the company’s agreement to extend an existing guardrail to prevent users from accessing certain restricted capabilities, according to two people familiar with the matter. This new safeguard ensures any attempts to unlock those capabilities will result in users being notified their request is blocked and their query will be processed by the less-advanced Opus 4.8 AI model. The change marks a significant step in resolving tensions between Anthropic and the administration, which had previously imposed export controls on the model.
The new safeguard extends the guardrail to requests related to 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 bypass a restriction on Fable 5 by asking the model to fix code rather than identify security issues. While cybersecurity experts generally do not find this behavior troubling, the administration’s awareness of it led to the imposition of export controls, effectively taking the model offline. The addition provides new insight into Commerce Secretary Howard Lutnick’s letter announcing the removal of restrictions on Anthropic’s Fable 5 and Mythos 5 AI models.
Lutnick’s letter highlighted Anthropic’s agreement to proactively detect and address security risks posed by the models. WIRED first obtained the letter and shared its details on Tuesday night. The Commerce Department also cleared Fable 5 for release after researchers at its Center for AI Standards and Innovation determined the model’s safeguards were sufficiently robust. Despite resolving its impasse with the Commerce Department, defense secretary Pete Hegseth has informed advisers there is no clear path to lift his February 28 order designating the company a supply chain risk.
Source: wired
