Journalism begins where hype ends

,,

I visualise a time when we will be to robots what dogs are to humans, and I'm rooting for the machines"

— Claude Shannon

Claude Mythos Won’t Stay Rare or in Safe Hands Forever, Warns Anthropic Frontier Red Team Head

Composite image showing the Claude Mythos logo on one side and Logan Graham, head of Anthropic's Frontier Red Team, on the other.
May 14, 2026 02:36 AM IST | Written by Vaibhav Jha

Putting a word of caution against the increasing capabilities of artificial intelligence (AI) models, Logan Graham, head of Frontier Red Team at Anthropic, said that models like Claude Mythos won’t stay rare or in safe hands forever.

On X, Graham informed that Claude Mythos Preview has helped find many thousands of “estimated” highly critical severity vulnerabilities in security systems of Anthropic’s partners under Project Glasswing.

“Within a year, Mythos will probably look quite dumb (relative to other new models). And others may release openly available or unguardrailed models of Mythos-level capabilities. We started Project Glasswing because capabilities like Mythos Preview’s won’t stay rare, or stay in careful hands,” said Graham.

Claude Mythos is a frontier AI model developed by Anthropic AI, claimed to have superior capabilities than existing AI models, thereby posing cybersecurity risks for institutions and businesses.

Anthropic has not released Mythos for general use and rather unveiled a preview version selectively available to tech giants and institutions under Project Glasswing, to “find and fix vulnerabilities or weaknesses in their foundational systems.”

 

Following the unveiling of Claude Mythos Preview model, several governments including U.S. U.K, India and Singapore have issued advisories to their respective banking institutions and businesses to overhaul their security management.

Graham, who heads the Frontier Red Team at Anthropic, informed that Claude Mythos has become the first AI model to have successfully passed UK’s AI Security Institute(AISI)’s cooling tower cyber ranges test. Additionally, an independent assessment by XBOW has also shown a step-change in autonomous cyber capabilities of AI models.

“I don’t share this to boost Mythos. In fact, this is not about Mythos. It’s about preparing for the coming world of models being better, faster, cheaper, and more creative than some of the best human experts at dual use capabilities,” added Graham.

Also Read: SEBI Flags Claude Mythos in Advisory, Asks Indian Markets to Level-up Defense

Author

  • Vaibhav Jha, editor and co-founder at AI FrontPage

    Vaibhav Jha is an Editor and Co-founder of AI FrontPage. In his decade long career in journalism, Vaibhav has reported for publications including The Indian Express, Hindustan Times, and The New York Times, covering the intersection of technology, policy, and society. Outside work, he’s usually trying to persuade people to watch Anurag Kashyap films.

    LinkedIn

Related Posts

LATEST NEWS