U.S. President Donald Trump stated he no longer considers artificial intelligence company Anthropic a national security threat, a change from his previous stance. The shift occurred after a dispute over foreign access to Anthropic's advanced AI models, Fable 5 and Mythos 5.
Last week, the U.S. Commerce Department issued a directive to Anthropic, invoking an export control that banned non-Americans from accessing the models, citing unspecified national security concerns. Anthropic responded by disabling access to both models for all users to comply with the directive. The company believes the action may be related to a bypass of the model's guardrails, though details were not provided.
President Trump described Anthropic CEO Dario Amodei's response as "very quickly" and "responsibly." Trump met with tech leaders, including Amodei, at a G7 summit and did not rule out using emergency powers under the Defense Production Act against Anthropic, though he expressed uncertainty about its necessity. An Anthropic spokesperson expressed gratitude for the administration's partnership in resolving the matter.
Security researchers have questioned the government's reasoning, suggesting the bypass described in a paper shared by Anthropic should not trigger an export control. Experts have called for the directive to be revoked, warning that restricting advanced cybersecurity capabilities is dangerous. Some commentators suggest the action may have been retaliatory or based on personal/political factors, setting a precedent for government control over software release.