Cybersecurity executives urge the Trump administration to ease restrictions on Anthropic AI models

A coalition of cybersecurity executives and experts, including representatives from Adobe and Nvidia, signed a letter Sunday calling on the Trump administration to reverse its export control directive on Anthropic’s newest AI models, Fable 5 and Mythos 5. The letter argues the restrictions could weaken U.S. cybersecurity defenses by limiting access to advanced tools that help identify and exploit software vulnerabilities. Anthropic announced Friday it had removed the models from public use to comply with the directive, stating the government’s actions were disproportionate to the identified risks. The letter, signed by over 100 professionals, acknowledges that Mythos models excel at finding software flaws but emphasizes they are not uniquely superior to other foundation or open-source models. It warns that removing these capabilities without justification risks ceding an advantage to adversaries, particularly China, whose AI models are reportedly only months behind U.S. leaders. The authors also highlight concerns that China may possess undisclosed, more advanced capabilities. The Commerce Department’s directive marks the U.S. government’s most significant effort yet to restrict access to cutting-edge AI, following President Donald Trump’s executive order last week. The order establishes a voluntary framework for AI developers to assess national security risks before public release, allowing up to a month for federal vetting. Anthropic had previously engaged with the White House over concerns about its models’ capabilities, including their potential to surpass human cybersecurity experts in identifying vulnerabilities. Tensions between Anthropic and the Trump administration have escalated, particularly after a contract dispute with the Pentagon. Defense Secretary Pete Hegseth labeled Anthropic a supply chain risk, a move the company has contested in federal court. Anthropic has insisted on safeguards to prevent its technology from being used in fully autonomous weapons or for surveillance of Americans, while the Pentagon has pushed for broader applicability. The letter’s authors urge the government to adopt a transparent, scientific process for evaluating AI risks, arguing that arbitrary restrictions could undermine U.S. innovation and security. The Commerce Department declined to comment on the matter as of Monday. Meanwhile, Anthropic released a limited version of its Fable model last week, while tightly controlling access to the more advanced Mythos model due to cybersecurity concerns.