Anthropic meets White House after U.S. blocks new AI models

Anthropic’s sudden White House meeting has become an early test of how far the United States is willing to go to use export-control power on frontier AI. The company said it was forced to block access to its newest models after receiving a government directive at 5:21 p.m. ET on June 12, and the fight now centers on whether that order was a one-off suspension or the outline of a new federal playbook for controlling advanced AI.

Anthropic said the directive required it to suspend access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national employees. To comply, Anthropic disabled the models for all customers worldwide. The company said the letter did not spell out a specific national-security concern, but that the government appeared to believe it had identified a way to bypass, or “jailbreak,” Fable 5.

Anthropic pushed back on that reading, saying its own review found only a small number of previously known, minor vulnerabilities. The company said other publicly available models could find the same weaknesses without needing a bypass. It also said Fable safeguards had been tested for thousands of hours before launch with the U.S. government, the UK AI Security Institute, private third parties and internal teams, and that no tester had found a universal jailbreak.

The dispute escalated over the weekend as senior Anthropic technical staff went to Washington to meet with White House officials, with virtual meetings already underway since Friday. A Monday session with Commerce Department officials was expected to include more documentation on the alleged issue. The timing matters because the government move landed as Anthropic was already taking an unusually cautious approach to its most advanced systems: the company had previously warned about Mythos’s hacking capabilities and kept it from wide release before releasing the public Fable version with cybersecurity safeguards.

The backlash has been swift. More than 50 cybersecurity leaders, including people from Nvidia and Adobe, urged the Trump administration on June 15 to lift the restrictions, arguing the ban would hamper defenders trying to find and fix software flaws and could weaken U.S. AI leadership. The episode puts a wider economic question in focus: whether Washington is prepared to treat advanced commercial AI models as strategic technology subject to export-style controls, with direct consequences for domestic companies competing globally. Anthropic is also confidentially filing for a U.S. initial public offering, raising the stakes for how regulators handle the next frontier model dispute.