More than 50 cybersecurity leaders from major U.S. companies including Nvidia and Adobe have asked the Trump administration to reverse restrictions on Anthropic's most powerful AI models, warning the bans are hurting efforts to defend against digital attacks. The letter, sent on Sunday, comes just days after Washington ordered Anthropic to suspend access to its Fable 5 and Mythos 5 models for any foreign nationals over national security concerns. The security experts argue that taking away these tools from defenders creates more risk than it prevents.
The restrictions stem from Friday's government decision to block foreign nationals from accessing Anthropic's latest models after the company had previously warned about the hacking capabilities of its Mythos model and withheld it from wide release. Anthropic then released a public version last week called Fable, which included what the company described as cybersecurity safeguards. The access ban now applies to both the Fable 5 and Mythos 5 models, despite the company's deployment of the commercial Fable version to hundreds of millions of people.
The letter from security leaders states that Anthropic's models weren't uniquely capable of finding security flaws and weaponizing exploits, with many rival models offering similar abilities. "This action has taken the best models away from defenders, created market uncertainty, and risked America's AI leadership without any real risk to justify it," the letter stated. The cybersecurity executives argued that removing access to Anthropic's latest models at a time when China is rapidly advancing its own AI capabilities is "dangerous." Anthropic itself said in a blog post that it disagrees that finding a narrow potential jailbreak should justify halting access to a commercial model deployed to hundreds of millions of users.
The restrictions underscore growing tensions between AI companies and Washington over balancing innovation with national security. According to the report, the curbs limit the cybersecurity industry's ability to find and fix software flaws at precisely the moment when other AI tools are making it easier for hackers to exploit vulnerabilities. CrowdStrike reported last week that China-linked hackers posed the biggest espionage threat to technology companies over the past year, adding urgency to the defenders' argument. Earlier this year, Trump directed U.S. agencies to stop working with Anthropic entirely and declared it a supply risk because of the company's reluctance to let its technology be used for mass surveillance and autonomous weapons.
Those tensions appear to be easing as Anthropic, now valued at $965 billion, prepares to go public. The timing of the letter suggests the security community sees the access restrictions as counterproductive—defenders need the same powerful tools that potential attackers might use to stay ahead of threats. If the ban stands, American cybersecurity teams could find themselves outgunned by adversaries using equally capable AI models from other sources, while the country's AI leadership position faces unnecessary risk from regulatory uncertainty.