Fable 5 & Mythos 5 Offline: Export Controls Hit AI Cybersecurity Models

For cybersecurity professionals, the directive to pull Fable 5 and Mythos 5 highlights the growing recognition that advanced AI models can be weaponized by foreign adversaries. Mythos 5 had already been restricted due to its rapid exploit creation capabilities, and now export controls aim to close loopholes that could let these tools fall into the wrong hands.

In a dramatic escalation of AI governance, Anthropic removed its two most advanced AI models—Fable 5 and Mythos 5—from public access late Friday afternoon, June 12, 2026, after receiving an urgent directive from the U.S. Commerce Department. The directive, issued under new export control protocols, orders the company to prevent use of the models by foreign nationals, marking the most significant federal intervention into cutting-edge AI technology to date. The action comes just 10 days after President Donald Trump signed an executive order establishing a voluntary framework for national security vetting of advanced AI systems, a timeline that suggests the administration is moving aggressively to assert control over frontier AI capabilities.

“

In a dramatic escalation of AI governance, Anthropic removed its two most advanced AI models—Fable 5 and Mythos 5—from public access late Friday afternoon, June 12, 2026, after receiving an urgent directive from the U.S.

Anthropic had widely released Fable 5 earlier that week, positioning it as a powerful but limited version of the even more advanced Mythos 5, which the company had kept under tight restrictions due to its rapid exploit creation capabilities—a clear cybersecurity concern. By taking both models offline, the government is effectively treating even the less restricted version as a potential national security risk. Anthropic’s public statement did not mince words: while acknowledging the government’s legitimate interest in blocking unsafe deployments, the company denounced the process as a “misunderstanding” that lacked transparency, fairness, and technical grounding. “We believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts,” the company said. “This action does not adhere to those principles.” The Commerce Department has remained silent, offering no immediate comment.

The implications extend far beyond one company. This is the first test of a new export-control regime that could fundamentally reshape the AI landscape. Until now, the U.S. has relied on export controls for hardware—like advanced semiconductor manufacturing equipment—but extending them to pure software and model weights represents uncharted territory. The move raises immediate legal questions: does the executive order provide sufficient statutory authority, or will Anthropic and other AI developers challenge it in court? The company’s strong statement signals that it is prepared to fight, arguing that the process was arbitrary and lacking the “technical facts” that should underpin such a decision.

The impact on the AI industry is chilling. Other frontier labs—OpenAI, Google DeepMind, Meta—will be watching closely, as their own model releases could be subject to similar sudden restrictions. Investors in private AI companies may begin pricing in a new regulatory risk premium, potentially slowing the breakneck pace of funding. For Anthropic itself, the shutdown of access to Fable 5 just days after its release disrupts enterprise partnerships and erodes confidence among customers who rely on the model’s capabilities. The company’s claim that Mythos 5 was already limited underscores the dilemma: even voluntarily imposed restrictions were not enough to satisfy the government’s new standards.