Key Points:
- President Donald Trump announced he no longer views artificial intelligence developer Anthropic or its CEO Dario Amodei as a national security threat.
- The dramatic shift in tone follows a high-profile working lunch at the G7 summit in France, where Amodei pitched a U.S.-led AI coalition.
- While Trump praised the company’s rapid compliance with recent export curbs, the government’s official restrictions on Anthropic’s top models remain active.
- The regulatory resolution is financially critical for Anthropic as the firm prepares for a highly anticipated October initial public offering.
United States President Donald Trump has dramatically softened his stance on artificial intelligence developer Anthropic PBC, declaring that he no longer views the company or its Chief Executive Officer, Dario Amodei, as a national security threat. The sudden shift in rhetoric represents a remarkable turn of events, coming just a week after the federal government enacted aggressive, emergency export-control restrictions against the company’s most advanced software systems. In a pre-taped media interview, the president credited the startup’s leadership for acting with extreme responsibility and cooperating fully with federal investigators to secure their systems from potential foreign exploitation.
When asked directly during the interview if he considered the San Francisco-based startup or its chief executive a threat to national security, Trump replied, “Well, not now. But a week ago, maybe.” The president emphasized that the regulatory standoff had reached a critical phase because the legal liabilities of non-compliance are exceptionally severe. He warned that individuals face immediate prison time for violating federal export-control directives. However, Trump praised Amodei’s rapid, compliant response to the administration’s emergency orders, signaling that negotiations to resolve the high-profile standoff are progressing positively.
The primary catalyst for this diplomatic breakthrough was a high-profile working lunch at the Group of Seven (G7) summit in Évian-les-Bains, France. The international summit provided Amodei with a rare, face-to-face venue to repair his damaged relationship with the White House. During the working lunch, which included several global heads of state and top technology executives, Amodei reportedly pitched President Trump and other Western leaders on a comprehensive, U.S.-led international artificial intelligence coalition. The proposal aims to unify Western computing power and software development to maintain a decisive technological lead over global adversaries.
This personal diplomacy follows a chaotic week that began when the Department of Commerce issued an emergency export-control directive. The federal order required Anthropic to obtain prior government approval before allowing any foreign national—including its own international technical staff—to access its newly released Claude Fable 5 and Mythos 5 models. Because the company’s cloud infrastructure could not reliably separate foreign and domestic users in real-time, Anthropic chose to abruptly take both models completely offline worldwide to ensure compliance, leaving thousands of corporate developers instantly locked out of its premier systems.
The federal government’s initial, aggressive intervention was triggered by specific intelligence reports regarding foreign espionage. U.S. national security officials reportedly became aware of a potential “jailbreak” vulnerability within Fable 5’s safety guardrails. At the same time, intelligence units discovered that a major South Korean telecommunications firm, which the administration suspected of maintaining close technological ties to Chinese state interests, was slated to receive early access to the highly capable Mythos 5 model. Because Mythos 5 possesses advanced abilities to identify software code vulnerabilities, Washington feared the technology could be exploited to launch devastating cyberattacks against domestic infrastructure.
The recent software clash represents the continuation of a long-running, highly public feud between the tech company and the current administration. In March, the Department of Defense issued a formal supply-chain risk designation that barred all federal agencies from using Anthropic’s products. That federal ban was triggered after contract negotiations broke down over Anthropic’s refusal to remove ethical safety guardrails from its military-facing products. The company’s leadership had refused to eliminate built-in restrictions related to the mass surveillance of citizens and the development of fully autonomous lethal weapons, leading Defense Secretary Pete Hegseth to publicly accuse the startup of duplicity.
While the president’s new comments are highly encouraging for the technology sector, industry analysts warn that the actual, formal policy has not yet changed. Despite the optimistic rhetoric, both the Commerce Department’s June 12 export-control order and the Pentagon’s March supply-chain risk designation remain formally in place and have not been rescinded. Until these regulatory barriers are officially lifted, Anthropic remains legally barred from distributing its most advanced models to international customers or utilizing its own foreign national employees for core software development.
During his media interview, the president also discussed his broader regulatory philosophy for the fast-growing technology sector, noting that he retains the power to invoke the Defense Production Act (DPA) to enforce compliance. He warned that he has the authority to use a wide variety of emergency powers if technology companies refuse to cooperate with national security directives. However, Trump indicated that such aggressive, heavy-handed measures would not be necessary if other AI developers follow Anthropic’s example of rapid cooperation, reiterating his belief that the massive benefits of artificial intelligence far outweigh its potential risks.
For Anthropic, securing a formal, permanent resolution to these regulatory disputes is an absolute matter of corporate survival. The startup recently filed confidentially for an initial public offering (IPO), with its planned October listing targeting a highly anticipated corporate valuation of $965 billion. Because the company’s financial projections and global expansion plans depend entirely on its ability to sell its advanced models to international enterprise clients, any prolonged regulatory ban would severely damage its public listing. The ongoing standoff proves that in the modern digital age, a tech firm’s ultimate valuation is determined not just by its laboratory breakthroughs, but by its alignment with the national security goals of Washington.
