Anthropic's Fable 5 and Mythos 5 models shut down after US government security order

At a glance:

• Anthropic completely shut down access to Fable 5 and Mythos 5 models following a June 12th US government order citing national security concerns.
• The shutdown affects all users globally, including Anthropic employees, after security research from Amazon reportedly identified potential cyberattack vulnerabilities.
• The incident intensifies arguments about over-reliance on US-based AI and provides ammunition for other nations seeking technological independence.

The Trump administration's action came swiftly and with limited explanation, ordering Anthropic to block access to its newest and most powerful AI models for all foreign nationals, both within and outside the United States. This unprecedented shutdown occurred just three days after Fable 5 and Mythos 5 launched on June 9th, when Anthropic claimed Fable 5's capabilities exceeded any model the company had previously made generally available. The restriction included Anthropic's own researchers, many of whom are foreign-born, effectively cutting off access even for the engineers who built these systems.

According to reports from The Wall Street Journal, the export control directive was triggered in part by cybersecurity research conducted by Amazon. The tech giant reportedly found ways to prompt Fable 5 into generating information that could be leveraged for cyberattacks. Following CEO Andy Jassy sharing these findings with the White House, the government moved quickly to impose the restrictions. Anthropic confirmed compliance but pushed back on the characterization of the vulnerabilities, arguing that evidence of potential jailbreaking was provided only verbally and that the discovered weaknesses were minor and available through other models like GPT 4.5.

The White House's concerns extended beyond immediate security vulnerabilities to include fears about Chinese access. Semafor reported that government officials worried a group linked to China may have accessed Mythos 5, which would pose serious national security risks including potential reverse engineering through model distillation techniques. While the Trump administration has not confirmed these specific concerns, White House adviser David Sacks addressed the situation on X, focusing on reported jailbreaking capabilities rather than Chinese access allegations. Anthropic has not responded to requests for comment, though a spokesperson told Semafor that Chinese government involvement was not discussed during export control conversations.

This sudden shutdown has reinforced longstanding arguments against over-reliance on US technology infrastructure, particularly in critical sectors. Politicians, governments, and companies already advocating for technological independence have found new momentum in their positions. The incident underscores how US government authority can abruptly restrict access to cutting-edge AI capabilities, affecting users worldwide and highlighting the geopolitical complexities of frontier AI development. For international organizations and foreign governments, the event serves as a stark reminder that access to US-developed AI remains subject to political and security considerations beyond commercial agreements.

Fable 5 represents Anthropic's latest advancement in its Claude model lineup, positioned as the company's most capable publicly available system. According to Anthropic, the model demonstrates exceptional performance in software engineering, knowledge work, and vision tasks, with its advantages becoming more pronounced as task complexity increases. Claude Mythos 5 shares the same underlying architecture but operates with certain safeguards relaxed in specific areas, reflecting Anthropic's ongoing balancing act between capability and safety controls. Both models were designed for broad commercial deployment, reaching hundreds of millions of users upon their initial release.

The broader implications extend to the global AI landscape, where the United States maintains significant influence over frontier model development and distribution. Other nations and technology leaders are increasingly scrutinizing their dependencies on US AI infrastructure, accelerating investments in domestic alternatives. This incident may reshape procurement decisions across government agencies and enterprise customers worldwide, potentially spurring increased funding for European, Chinese, and other regional AI development initiatives. The question of who controls access to powerful AI systems is becoming as important as the capabilities themselves.

Moving forward, the incident raises questions about transparency in government-AI company interactions and the adequacy of current frameworks governing international AI access. Industry observers will be watching how Anthropic navigates this situation and whether other AI developers face similar pressures. The intersection of national security considerations and commercial AI deployment continues to evolve, with this case establishing new precedents for how governments can intervene in private sector AI development. Companies developing frontier AI models must now consider not just technical and ethical challenges, but also the complex regulatory and geopolitical landscape that can abruptly reshape their market strategies.