Anthropic has suspended access to its advanced Claude Fable 5 and Mythos 5 AI models after a US government security directive raised concerns over potential vulnerabilities.
 |
| Anthropic has disabled its latest AI models globally after a US government directive citing national security concerns and possible jailbreak vulnerabilities. Image: CH |
Tech Desk — June 13, 2026:
Anthropic’s decision to suspend access to its most advanced AI models marks one of the clearest signs yet that artificial intelligence development is entering a more tightly regulated and politically sensitive phase.
The company disabled Claude Fable 5 and Claude Mythos 5 after receiving a US government directive citing national security concerns. The order reportedly required restricting access to foreign nationals, including non-US employees, forcing Anthropic to shut down the models globally in order to comply.
The move is significant not only because of the technology involved, but because of how quickly it unfolded. Fable 5 had only been publicly available for a short period before access was abruptly withdrawn, highlighting how regulatory intervention can now reshape AI deployment in real time.
At the center of the issue is concern over potential “jailbreaking” techniques—methods used to bypass built-in safety restrictions in AI systems. While the government did not publicly detail the specific threat, Anthropic said it understood the concern to relate to possible misuse of the model’s advanced capabilities.
The company, however, pushed back on the severity of the issue. It stated that internal reviews found only limited and previously known vulnerabilities, arguing that similar weaknesses exist across other widely available AI systems. This disagreement underscores a growing tension between AI developers and regulators over how risks should be assessed and defined.
The suspension also reflects the dual-use nature of modern AI. Models like Fable 5 and Mythos 5 are designed to be highly capable, including in areas such as cybersecurity and complex reasoning. But those same capabilities raise concerns that they could be repurposed for malicious use if safety barriers are bypassed.
This is not the first time advanced AI systems have faced restrictions, but the scope of this action is notable. By requiring global shutdown rather than regional limits, the directive effectively demonstrated how national security considerations can override commercial rollout strategies.
The case also highlights the growing geopolitical dimension of AI development. The requirement to restrict access for foreign nationals points to increasing concerns about how cutting-edge models might be used across borders, particularly in sensitive technological or security contexts.
Anthropic’s response suggests a broader industry debate is intensifying. While companies are investing heavily in safety research and guardrails, they also argue that over-regulation or overly cautious interpretations of risk could slow innovation and lead to fragmented global access to AI tools.
Despite the suspension of Fable 5 and Mythos 5, Anthropic confirmed that its other AI systems remain available. Still, the incident signals that even limited safety concerns—whether fully validated or not—can now lead to sweeping operational consequences.
The situation also adds to ongoing legal and policy friction between AI companies and US authorities over how advanced models should be controlled, especially when they are capable of performing high-level reasoning and potentially sensitive technical tasks.
More broadly, the episode reflects a turning point in AI governance. As models become more powerful, decisions about access, safety, and national security are increasingly being made not just by companies, but by governments asserting direct influence over deployment.
The suspension of Anthropic’s flagship models therefore stands as a warning that the era of unrestricted rollout for frontier AI systems may be narrowing, replaced by a more controlled and security-conscious regulatory environment.