Trump backs off Anthropic national security warning after AI access dispute

President Donald Trump has walked back a national-security warning against Anthropic just days after treating the company’s AI models as a threat, a sharp reversal that now looks less like a settled judgment than a test of White House leverage over frontier AI. The change followed a dispute over foreign access to Anthropic’s most advanced systems, Fable 5 and Mythos 5, after the government ordered the company to suspend access for foreign nationals.

Anthropic said it received the directive on June 12, 2026, at 5:21 p.m. ET, and that it applied to any foreign national inside or outside the United States, including foreign national employees. The company said officials did not spell out the national-security concern, though it believed the government was reacting to a reported jailbreak method for Fable 5. Anthropic said it had already worked for thousands of hours on red-teaming before launch with the U.S. government, the UK AISI, private third-party organizations and internal teams.

Trump’s new tone matters because it suggests the administration is still drawing the line between innovation policy and security enforcement in real time. The president said Anthropic’s leadership responded “very quickly” and “responsibly,” and he left open the possibility of using emergency authorities under the Defense Production Act, though he said he was not sure he needed to do so. That combination of retreat and threat leaves industry with a clear message: access to advanced AI can be tightened fast if the White House decides national security is in play.

The episode also exposed how closely AI firms are now tied to the machinery of government. POLITICO reported that the White House and Anthropic are discussing a framework to rate the severity of security flaws in new AI models, with talks led by Sarah Heck, Anthropic’s head of public policy, and co-founder Tom Brown. The same report said the dispute followed a frantic 24-hour stretch involving Dario Amodei, Treasury Secretary Scott Bessent, White House Cyber Director Sean Cairncross, White House chief of staff Susie Wiles and Amazon CEO Andy Jassy.

Trump’s broader AI posture has not been purely restrictive. On June 2, 2026, he signed an executive order aimed at advancing AI innovation while strengthening cybersecurity and protecting critical infrastructure, but it did not create mandatory licensing, pre-clearance or permitting for AI models. Anthropic launched Fable 5 on June 9 and said it was the first model to break 90% on its core analytics benchmark for complex, long-running tasks, which helps explain why the company’s access became strategically sensitive so quickly.

For investors, competitors and foreign customers, the bigger lesson is not just that Trump changed his mind. It is that AI policy in Washington can pivot from growth language to security intervention in a matter of days, and that once a model is framed as a national-security issue, the rules around access, exports and oversight can change just as fast.