2 Top AI Models Disabled by Anthropic After US Jailbreak Order

For AI researchers and developers, the U.S. government’s reliance on verbal evidence of a narrow jailbreak to ban commercial models sets a troubling precedent. It signals that any advanced AI model might be yanked from the market on thin grounds, undermining the scientific rigor needed for safety evaluations and threatening the open development ecosystem.

Anthropic's announcement on June 12, 2026, that it will 'abruptly disable' its most advanced AI models—Fable 5 and Mythos 5—for all users marks a dramatic new chapter in U.S. technology export controls. The company said it received a directive from the U.S. government under export control laws to suspend access to these models for all foreign nationals, citing national security concerns. The government's case rests on a claim that there exists a method of 'jailbreaking' a specific safeguard in Fable 5, potentially allowing the model to be used to identify software vulnerabilities. Anthropic contests the order, asserting it was given only 'verbal evidence of a potential narrow, non-universal jailbreak' and that recalling a commercial model deployed to hundreds of millions of people is disproportionate.

“

Anthropic's announcement on June 12, 2026, that it will 'abruptly disable' its most advanced AI models—Fable 5 and Mythos 5—for all users marks a dramatic new chapter in U.S.

This action represents a significant escalation in the U.S. government's approach to AI regulation. Historically, export controls have targeted the physical chips and tools that power AI—such as those restricted under the CHIPS Act and related executive orders. By turning its focus to the AI models themselves, the government is setting a precedent that could reshape the global AI landscape. The order also comes just as Anthropic had been seeking to mend relations with the Trump administration after a public rupture earlier in 2026, when the company refused to allow the U.S. military to use its AI for domestic surveillance and fully autonomous weapons systems. That refusal led the government to place Anthropic on a supply chain blacklist, slated to take effect later this year. The latest directive may be viewed as an extension of that conflict, though the government frames it as a direct response to a jailbreak threat.

For Anthropic, the consequences are immediate and severe. Disabling Fable 5 and Mythos 5 for all users—not just foreign nationals—will disrupt services for hundreds of millions of users and could undermine its IPO ambitions. The company’s valuation and investor confidence hinge on its ability to deploy cutting-edge AI globally; a forced withdrawal from foreign markets, even temporarily, could shrink its revenue streams and raise existential questions about its business model. The lack of due process—receiving only verbal evidence without formal technical documentation—also raises legal questions about the scope of executive authority under the International Emergency Economic Powers Act (IEEPA) or other export control statutes. If Anthropic challenges the order in court, it could become a landmark case defining the boundaries of AI export controls.

The broader AI industry is watching closely. Other major AI developers—OpenAI, Google DeepMind, Meta—may now face similar restrictions if the government expands its rationale. The vague standard of a potential jailbreak that enables vulnerability discovery could be applied to any advanced model, effectively granting the government broad discretion to block model access. This unpredictability could chill investment in frontier AI research, particularly for startups that lack the legal resources to fight such orders. Venture capital flows into AI startups might decelerate if regulatory risk is perceived as too high. Moreover, the action could accelerate the balkanization of AI: foreign nations may redouble efforts to develop indigenous AI capabilities, reducing U.S. influence and potentially leading to a race to the bottom in safety standards.