US President Donald Trump has softened his stance towards Anthropic, one of America’s leading artificial intelligence developers. In an interview with“The Axios Show”, he admitted that just a week earlier he might have viewed the company or its CEO, Dario Amodei, as a potential threat to national security, but that he no longer shares that view.
The statement comes at a time of tension between the administration and Anthropic over the Fable 5 and Mythos 5 models. A few days ago, the US government imposed export restrictions on them, banning access to foreign users without the appropriate licences. In response, the company temporarily disabled access to both models for all customers, arguing that this was the quickest way to ensure compliance with the authorities’ decision.
According to US media reports, the Trump administration fears that Anthropic’s advanced systems could be used to detect security vulnerabilities or for other applications of strategic importance. There are also reports of potential ways to circumvent the models’ security mechanisms, although the company emphasises that the evidence presented to it related to limited scenarios and did not indicate a widespread threat.
Trump, however, assessed that Anthropic had responded to the government’s guidelines “very quickly” and “responsibly”. At the same time, he did not rule out the use of emergency powers under the Defence Production Act, although he noted that he did not see any need for this at present.
Representatives of Anthropic have declared their readiness to continue cooperating with the administration. The company emphasises that the shared goal remains the protection of critical infrastructure and maintaining the United States’ position as a leader in the development of artificial intelligence.
The dispute highlights that advanced AI models are increasingly becoming not just a technological product, but also an element of national security policy. For the industry, this means a growing role for regulation and greater government influence over how the most advanced artificial intelligence systems are made available.
