The Trump administration has forced Anthropic to pull its latest artificial intelligence models offline, marking a significant escalation in state control over advanced technology and revealing the government's role in securing military and corporate interests. This directive came after Defense Secretary Pete Hegseth sought to designate Anthropic as a supply chain risk following a Pentagon contract dispute, a move Anthropic has challenged in two federal courts. The government's actions have raised concerns among cybersecurity executives and policy analysts that a precedent is being set for direct intervention in deployed AI models through export control authorities.
Anthropic announced on Friday, 4 days ago, that it complied with a U.S. Commerce Department directive that barred non-Americans, including Anthropic’s own employees, from accessing its Fable 5 and Mythos 5 models. The company stated it did not believe the government’s steps were warranted by the security concern flagged. Anthropic had previously limited the use of some of its latest technology to select customers due to its capacity to surpass human cybersecurity experts in identifying and exploiting computer vulnerabilities. The San Francisco-based company had discussed these capabilities with the White House prior to the directive.
The State's Grip on Technology
The Commerce Department's directive represents the U.S. government’s most significant step yet to restrict access to advanced AI models. Anthropic had widely released Fable about 1 week ago, a limited version of the more advanced Mythos, access to which had been tightly controlled due to cybersecurity fears. TechCrunch reported that the government’s enforcement letter effectively forced Anthropic to pull its models offline just before the weekend, demonstrating that the AI industry is not immune to state interference. TechCrunch further noted that the move served as a warning to the wider tech industry: comply, or the government can shut companies and their products down. The government’s action appeared to require no court approval, and the letter itself has not been made public.
Tensions between the Trump administration and Anthropic have been high. Anthropic had sought to implement guardrails on AI development to minimize potential risks and maximize its economic and national security benefits for the U.S. However, following a contract dispute, Defense Secretary Pete Hegseth attempted to declare Anthropic a supply chain risk, an unprecedented move against a U.S. company. Anthropic had sought assurance that the Pentagon would not use its technology in fully autonomous weapons or for the surveillance of Americans. Hegseth, representing the state’s military apparatus, insisted that the company must allow for any uses the Pentagon deemed lawful, revealing the state's demand for unfettered access to advanced technology for its coercive functions.
Capital's Demands and "National Security"
President Donald Trump signed an executive order 10 days ago to establish a framework for the federal government to vet national security risks of advanced AI systems for up to a month before their public release. This order, which stated that participation by AI developers would be voluntary, offers a symbolic concession while preserving the state's ultimate authority to define and control technological development under the guise of "national security." The Washington Post reported that Amazon CEO Andy Jassy first raised concerns about a Fable jailbreak with the White House, suggesting that Amazon’s strategic partnership with Anthropic was becoming more complex. This highlights the role of inter-corporate competition and influence in shaping state intervention.
More than 100 cybersecurity experts and leaders from companies including Adobe and Nvidia sent a letter asking the U.S. government to lift the export control directives. The letter argued that Anthropic’s Mythos models, while "quite good" at finding software flaws, are "not uniquely good at these tasks," and that many signatories regularly use other foundation and open-source models for security audits and training. These experts warned that taking away the best cyber defense capabilities "without a good reason" was dangerous, especially as "America’s adversaries are rapidly advancing," with China’s models "only months behind the best American models." This framing underscores the competitive drive between national capitals, where technological supremacy is a key battleground.
Industry Pushback and Its Limits
Katie Moussouris, a cybersecurity veteran and researcher who founded Luta Security, stated in a blog post that the bypass described by security researchers "should never have triggered an export control." Moussouris argued that the behavior described "cannot meaningfully be fixed, and any attempt would only weaken the model for defense." She, along with dozens of other top security researchers, called on the Trump administration to revoke the order, deeming the move to pull advanced cybersecurity capabilities from network defenders in the U.S. "dangerous."
Justin Hendrix, editor of Tech Policy Press, noted that the Trump administration’s move is "likely to raise alarms in foreign capitals about the reliability of American AI for critical applications." Hendrix added that the message conveyed is that AI companies in the United States cannot be trusted to operate without interference from the U.S. government. Joe Khawam of the Law Reform Institute indicated that the administration is likely to enforce export controls on AI models based on what they can say rather than where they are used, further solidifying the state's power to dictate the terms of technological development and deployment, ultimately serving the interests of accumulated wealth and its military-industrial apparatus.