On a Friday, Anthropic received a United States export control directive that mandated the suspension of access to its Mythos 5 and Fable 5 AI models for any foreign national, both within and outside the USA. The company concluded that the only feasible way to comply with this directive was to deactivate these AI models entirely. In an effort to reverse this decision, Anthropic representatives traveled to Washington, DC, with the aim of persuading President Donald Trump to reconsider the export control directive.
Claude Mythos 5 and Fable 5 are built upon the same foundational technology as Anthropic’s Mythos Preview. Mythos 5 was made available to a select group of government agencies and specific companies. In contrast, Fable 5 was assessed as safe for general public use and incorporated additional safeguards. However, a report surfaced suggesting that these safeguards on Fable 5 may not have been effective.
The Trump administration issued Anthropic a 90-minute ultimatum, demanding that the company cease all access to Mythos 5 and Fable 5. The directive stated that if Anthropic failed to comply, the government would impose export controls through the authority of the US Commerce Department. Within fifteen minutes of the initial call from the administration, Anthropic executives were engaged in discussions with the White House. Approximately one hour and fifteen minutes after the initial contact, CEO Dario Amodei joined these discussions. Amodei communicated directly with US Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, and National Cyber Director Sean Cairncross.
Anthropic communicated that it believed the government had become aware of a method to bypass or “jailbreak” Fable 5. The company described this jailbreak as a “potential narrow, non-universal” vulnerability that had been “shared with the government” by an unspecified entity. Anthropic further stated that this jailbreaking behavior was not unique to Fable 5 and was widely available from other AI models, including OpenAI’s GPT-5.5. Semafor reported that the US government’s concerns stemmed from the possibility that a group linked to China had accessed the technology. These concerns reportedly originated weeks prior and related to a large global telecommunications company that had initially been cleared for access to Mythos Preview.
David Sacks commented on an unnamed entity that had reportedly brought the issue to the government’s attention, describing it as “a highly credible trusted partner of both Anthropic and the USG who was testing Fable.” Some reports suggest that Amazon CEO Andy Jassy alerted the US government to concerns after Amazon researchers conducted red-teaming exercises on Fable 5. These findings were explicitly mentioned in conversations with the US government. According to a source, everything detailed in the Amazon research could be replicated using OpenAI’s GPT-5.5. Independent red-teamers, however, expressed that they were impressed by the level of protection offered by Fable 5.
Anthropic sent employees to Washington, DC, including Dave Orr, Logan Graham, and Nicholas Carlini, to address the situation. Axios reported that Anthropic had made repeated communication errors with the administration. For Anthropic, the timing of these events was particularly challenging, as the company had been anticipating Mythos to aid in its recovery from previous disagreements with the US Department of Defense. A source familiar with the negotiations indicated that Anthropic had pre-briefed the administration on Fable 5, and the US Department of Commerce had conducted pre-deployment testing without raising initial concerns. This source also noted that Anthropic had been collaborating closely with government agencies since the release of Mythos Preview.
While the Trump administration initially adopted a less interventionist stance on AI safety, its position has become more uncertain, particularly amid concerns about falling behind China in the AI race. Leading cybersecurity experts have warned that sidelining Mythos 5 and Fable 5 could provide China with a significant advantage in artificial intelligence. The directive from President Trump has intensified international calls for alternatives to American AI systems, effectively halting the progress of a major US AI company’s new flagship model. On Sunday, a public letter signed by executives from the tech and cybersecurity sectors advocated for the repeal of restrictions on Fable 5. The letter stated that if regulations were to be implemented, they should be based on “scientific evaluations developed with input from industry and academia.”