AI Model Availability
Anthropic's Fable 5 AI model has returned online for users following the Trump administration's decision to lift export controls, as reported by Axios. This move marks a significant policy change affecting AI technology's availability. Fable 5 is now accessible to all customers, with the caveat that queries deemed security or safety risks may be redirected to less powerful models. Customers will generally use Fable outside of subscription plans, paying per token used. Until July 7, subscribers can utilize Fable for up to half of their included data usage, though it consumes tokens more rapidly than other models.
Collaboration with Government
In a blog post, Anthropic detailed its collaboration with the US government, which included giving early access to frontier models, sharing information on new jailbreak methods, and joint research to advance AI evaluation. In a letter viewed by Reuters and The New York Times, Commerce Secretary Howard Lutnick confirmed that Anthropic no longer needs a license for exports or in-country transfers of its Claude Mythos and Claude Fable AI models, recognizing the steps taken to address risks.
Security Concerns and Precautions
The relaunch follows weeks of testing after Amazon researchers discovered a jailbreak technique that manipulated Fable 5 into producing code demonstrating how software vulnerabilities could be exploited. Anthropic stated that this jailbreak method is now blocked in over 99 percent of cases, with an improved safety classifier targeting and blocking the behavior. Users are notified if a request is blocked, and it is then sent to a less advanced model, Opus 4.8.
Future of AI Regulation
Anthropic CEO Dario Amodei called on Congress to pass legislation quickly, arguing that political institutions move too slowly relative to AI development and risk national strategic consequences if they cannot keep pace. Anthropic is working with Amazon, Microsoft, Google, and other partners in the Glasswing program—which allows cybersecurity researchers at trusted companies to access Mythos for defensive purposes—to draft a consensus framework for assessing the severity of AI jailbreaks and appropriate developer responses.
Global Coordination on AI Risks
Anthropic views its deepened partnership with the government as a template for global coordination on AI risks and benefits, urging Congress to pass laws to ensure all frontier model developers adhere to the same standards. Anthropic has accused Alibaba, a Chinese AI firm, of cloning attacks and has urged Congress to pass laws punishing Chinese firms found stealing US firms' work.
Conclusion
The relaunch of Anthropic's Fable 5 and the collaboration with the government highlight the evolving landscape of AI regulation and the need for a coordinated approach to managing the risks and benefits of advanced AI models. As the AI industry continues to grow, the role of government in ensuring safety and ethical use becomes increasingly critical.
