Cybersecurity leaders at major U.S. technology firms, including Nvidia and Adobe, have urged the Trump administration to lift restrictions on Anthropic’s most advanced artificial intelligence models, arguing that the limits could undermine efforts to detect and prevent cyberattacks.
The appeal follows a U.S. government decision on Friday ordering Anthropic to suspend foreign access to its Fable 5 and Mythos 5 models over national security concerns. The move has triggered pushback from more than 50 security executives who say the restrictions could slow the industry’s ability to identify software vulnerabilities at a time when AI tools are increasingly being used in cyberattacks.
In a letter circulated Sunday, the group argued that Anthropic’s systems are not uniquely capable of discovering or exploiting security flaws, noting that comparable capabilities exist in rival models, including China’s Kimi 2.7.
Joshua Saxe, chief technology officer at AI security firm Abundant Security and a signatory to the letter, told Reuters that Mythos represented only an incremental improvement over existing tools. “Mythos is almost definitely the best model right now for finding security bugs and codes, but it is like an incremental advance over other models that are already open,” he said.
Anthropic recently released a public version of its Fable model with cybersecurity safeguards after previously limiting access to its more powerful systems due to potential misuse risks. The company said it had withheld broader deployment of Mythos over concerns about hacking capabilities.
Senior Anthropic executives are expected to meet U.S. Commerce Department officials in Washington on Monday, according to a Trump administration official cited by Reuters.
The company has also argued that concerns about so-called “jailbreaking” techniques used to bypass safeguards do not justify restricting access to tools used by hundreds of millions of people. The industry letter echoed that view, warning that pulling back capabilities could be counterproductive as Chinese open-source models rapidly advance.
Alex Stamos, chief product officer at Corridor and another signatory, said regulation should be consistent and evidence-based. “This is an overreaction by the government,” he said, adding that there is disagreement over the severity of the technical findings that prompted the restrictions.
The debate comes as cybersecurity firm CrowdStrike recently warned that China-linked hackers remain the most significant espionage threat targeting technology companies.
Anthropic, valued at about $965 billion and preparing for a public listing, has previously clashed with U.S. authorities over national security concerns related to its AI systems. Earlier this year, the administration directed federal agencies to halt work with the company, citing risks tied to surveillance and autonomous weapons applications.
