Anthropic has received clearance from the US government to re-release its Mythos 5 and Fable 5 AI models. The move follows a letter in which the company “agreed to proactively detect and address security risks” associated with the technology, according to the Financial Times.
The agreement marks a significant moment for AI governance, blending federal oversight with private-sector development. A source told the Financial Times that Anthropic developed a dedicated “safeguard” for the models, though details of that mechanism remain undisclosed.
The re-release brings two previously suspended models back to market. The Financial Times reported that the US government's letter frames this as a conditional approval, with Anthropic now responsible for ongoing threat monitoring.
What comes next hinges on how effectively Anthropic's safeguard performs in practice. Developers and enterprise clients will now weigh the models' capabilities against the new regulatory guardrails, while competitors watch for precedent-setting enforcement.
No independent verification of the safeguard's efficacy has been published, and some critics question whether voluntary agreements can keep pace with rapidly evolving AI risks.