Anthropic has won government approval to restore its Claude Fable 5 and Mythos 5 AI models after a June ban over jailbreak risks. The company will relaunch Fable 5 globally while Mythos 5 rolls out to pre-approved organizations first, matching restrictions the Trump administration placed on OpenAI's GPT-5.6. Anthropic has installed new safety measures that block the flagged jailbreak technique in over 99% of cases and committed to closer government coordination on future model releases.
Summaries like this, in your inbox every morning.
Sign up free →What happened
The Department of Commerce lifted export controls on Anthropic's Claude Fable 5 and Mythos 5 models after weeks of negotiation with the Trump administration. Anthropic will begin restoring access to Fable 5 globally tomorrow, and will re-enable access on AWS, Google Cloud, and Microsoft Foundry soon, though with no set timeline. Mythos 5 is being restored only to a pre-approved list of organizations initially, similar to the rollout restrictions the Trump administration placed on OpenAI's GPT-5.6.
Why it matters
Anthropic sidelined both models in early June after a government export control directive citing concerns over potential jailbreaks—a major disruption as the company was preparing for an IPO and had just spent a week promoting these models. The restoration removes this operational constraint, though Anthropic remains under tighter government oversight, which may affect how quickly the company can scale international access and deploy new capabilities. The controls also signal that AI model security will remain a flashpoint between tech companies and U.S. regulators.
What to watch
Anthropic deployed an improved safety classifier that blocks the specific jailbreak technique flagged by Amazon researchers in over 99% of cases, and users whose requests are blocked will be rerouted to Opus 4.8 instead. The company has also committed to pre-release government access for models affecting national security, 24/7 monitoring of jailbreak submissions, and a new HackerOne program for researchers—steps that reshape how Anthropic will coordinate with government before releasing frontier models.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
