Anthropic halts Fable 5 access after U.S. export-control order tied to national security
Anthropic has suspended access to its newest models, Fable 5 and Mythos 5, after receiving a U.S. government export-control directive that the company says is rooted in national-security concerns.
In a statement published Friday, Anthropic said it received the directive at 5:21 p.m. ET instructing the firm to block access to Fable 5 and Mythos 5 for all foreign nationals — whether they are inside or outside the United States — and that the order also applied to foreign-national employees. To comply, Anthropic disabled the two models for all users. The company emphasized that its other models, including Opus 4.8, remain available and are not affected.
“We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users,” Anthropic said.
The shutdown comes just days after Anthropic rolled out Fable 5 as a generally available Mythos-class model and released Mythos 5 to a limited group of approved cybersecurity and infrastructure customers. Anthropic has positioned Fable 5 as a higher-capacity model capable of longer, more complex tasks and strong performance across software engineering, scientific research, finance, vision, memory, and knowledge work.
Anthropic said officials provided only verbal evidence pointing to a “narrow, non-universal” jailbreak technique that might bypass some of Fable 5’s safeguards. According to the company, the purported method involves asking the model to analyze a specific codebase and identify or repair software vulnerabilities — a targeted exploit rather than a broad removal of safety protections. Anthropic argues that a narrow jailbreak is materially different from a universal one and says recalling a commercial model over such a finding would, if broadly applied, effectively halt new model deployments across the frontier-AI industry.
The company says it is cooperating with authorities, believes the directive may stem from a misunderstanding, and is working to restore access as quickly as possible.
Why crypto stakeholders should care
The intervention highlights growing regulatory scrutiny of frontier AI and has ripple effects for crypto and web3 projects that increasingly rely on advanced models for code audits, security tooling, smart-contract analysis, and on-chain/off-chain automation. Restricting access to high-capacity models for foreign developers could complicate global security reviews and tooling that many crypto teams depend on.
Meanwhile, Anthropic is continuing to scale its compute and infrastructure plans. Private credit firms Blackstone and Apollo Global Management are syndicating roughly $36 billion in financing to back Anthropic’s next phase of infrastructure spending, Reuters reported. Anthropic plans to use the funds to acquire custom tensor-processing units (TPUs) from Google — chips that are backed by Broadcom technology and will be leased to the company for AI operations. That move, and the size of the financing, underscores how institutional capital is treating large-scale AI compute similarly to major infrastructure investments seen in crypto mining and hosting.
Policy push and broader stakes
Anthropic has also been pressing governments to establish stronger rules for frontier AI systems as model capabilities accelerate. Its policy proposals call for measures addressing dangerous deployments, independent evaluations, cybersecurity safeguards, and economic support for workers affected by automation. Those recommendations arrive as Anthropic now finds itself at the center of one of the most significant government interventions involving a newly released frontier AI model — a development that will shape debates about how to balance innovation, national security, and global access to cutting-edge AI.
Read more AI-generated news on: undefined/news
