Key Takeaways
- Export restrictions on Anthropic’s Fable 5 and Mythos 5 models were removed by U.S. authorities on June 30
- The ban began June 12 following discovery of a jailbreak exploiting security weaknesses
- Anthropic blocked all user access due to inability to confirm user nationalities instantly
- The company committed to sharing intelligence with authorities and providing early government access to new models
- Rising competition from Asian AI systems like Fugu and Tulongfeng influenced the decision to lift restrictions
Anthropic has regained the ability to offer its most sophisticated AI systems to the public following the U.S. government’s decision to withdraw export restrictions that had sidelined them throughout most of June.
Fable 5 became available worldwide again on July 1 through Anthropic’s service platforms. Mythos 5, which operates on identical underlying technology but features relaxed safety parameters, is now accessible to a select set of U.S.-based entities that secured governmental authorization on June 26.
The Reason Behind the Shutdown
The limitations commenced June 12 when federal authorities placed both AI systems on their export-controlled technology registry. This classification prohibited distribution to individuals of foreign nationality without explicit clearance.
The catalyst was a security vulnerability. Researchers from Amazon uncovered a jailbreak technique — a workaround designed to circumvent the model’s protective measures — that enabled Fable 5 to detect software security flaws and, in one instance, generate executable code demonstrating potential exploitation methods.
Since the directive was implemented without delay and Anthropic lacked systems to authenticate user citizenship on the fly, the organization chose to halt access completely rather than potentially violate federal regulations.
Security professionals questioned the necessity of the restrictions early on. They pointed out that Anthropic had already committed to resolving many of these security challenges on a voluntary basis, well ahead of any export regulation being established. Some observers suggested the prohibition served more as a negotiating tool than a legitimate security intervention, potentially connected to Anthropic leadership’s public disapproval of the Trump administration’s AI strategy.
Terms of Anthropic’s Commitment
Commerce Secretary Howard Lutnick announced that Anthropic accepted terms to actively identify and mitigate security vulnerabilities, collaborate with federal entities on guidelines for subsequent model deployments, and alert U.S. officials about any malicious usage patterns.
Anthropic is also expanding its partnership with government organizations. This encompasses granting authorized agencies preliminary access to cutting-edge models prior to public availability and exchanging threat intelligence data.
The organization is developing a standardized assessment system alongside Amazon, Microsoft, and Google to evaluate the severity level of specific jailbreak methods. These commitments align with a June 2 executive directive focused on AI security protocols.
International Competition Influenced the Decision
AI developers in Asia have started introducing systems that rival Mythos-level performance, such as Fugu and Tulongfeng. This competitive landscape created challenges for U.S. policymakers, undermining the rationale for restricting American AI technologies from international markets.
In recent days, Lutnick also authorized Mythos distribution to carefully vetted White House-sanctioned clients. OpenAI’s most recent systems followed a comparable distribution pattern, reaching only a curated list of approved institutions rather than being publicly available.
Anthropic remains a private company, though speculators have wagered on its valuation through a pre-IPO derivative instrument on the blockchain-based platform Hyperliquid. That derivative instrument declined approximately 3.7% when access was restricted in June.
The Trump administration’s evolving approach to AI regulation has created uncertainty throughout the sector regarding the framework that will shape future model deployments.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
