A group of leading cybersecurity experts is pressing the Trump administration to relax strict limits on Anthropic’s Mythos AI model. Former Facebook chief security officer Alex Stamos is leading the effort and warns the rules could make the internet weaker, not safer.
The administration recently ordered Anthropic to cut off access to Mythos 5 and Fable 5 for all foreign nationals. Anthropic said it took the systems offline “to comply with a directive from the Trump administration to prevent their use by foreign nationals,” calling it a “misunderstanding” it hopes to resolve quickly.
Officials acted after months of concern about Mythos’ extreme hacking abilities. In testing, the model reportedly found serious security flaws across “every major operating system and web browser” and could chain those weaknesses into end-to-end cyberattacks. Because of those results, Anthropic skipped a public launch and instead shared the model with about 40 large tech and security companies in a limited program called Project Glasswing.
What Alex Stamos and Security Leaders Are Arguing
In a letter highlighted by Axios, Stamos and other experts say Mythos is one of the few tools that lets defenders move faster than attackers. They say it can scan huge codebases, find zero-day flaws, and stress-test critical systems at a scale humans cannot match. Stamos warns, “We only have something like six months before the open-weight models catch up to the foundation models in bug finding,” so similar tools will likely appear soon.
The signers argue that strict limits on Mythos will not stop hostile states or criminal gangs from building similar AI. Instead, they say the United States should allow its agencies and major companies to safely use the strongest defensive AI available, with controlled access and independent risk tests, rather than broad bans that also affect security teams.
The Mythos fight comes as the Trump administration rewrites Washington’s approach to advanced AI. After canceling an earlier AI order, the White House is weighing a new system that would give officials a short review window for the most capable models before release, with special paths for cyber tools that protect critical infrastructure while still limiting offensive features.
Anthropic has sent senior engineers to meet Trump officials and argue that Mythos, if handled carefully, is more useful for defense than attack. Stamos and his peers say each week of tight restrictions leaves defenders weaker while attackers race to build their own AI hacking engines.
READ MORE: NEAR Price Prediction: Analysts Eye $4 as AI Tokens Explode Higher
