Anthropic Launches Mythos AI Security Effort With Apple and Other Tech Giants

Anthropic has launched a new cybersecurity effort called Project Glasswing, and Apple is one of the major technology companies involved as the company tests a powerful new AI model designed to find serious software flaws before attackers can use them. The project matters because Anthropic says its unreleased Claude Mythos Preview model has already uncovered "thousands of high-severity vulnerabilities," including flaws in every major operating system and web browser, which shows how quickly AI is changing the security landscape.

In a new announcement, Anthropic said Project Glasswing brings together Apple, Amazon Web Services, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks to help secure critical software. Along with those launch partners, more than 40 additional organizations that build or maintain important software systems now have access to the Mythos Preview model so they can scan their own code and fix weak points before the model's capabilities spread more widely.

Project Glasswing

Anthropic says the goal is simple: put advanced AI to work for defense before the same level of capability becomes common in the wrong hands. That urgency runs throughout the company's message, especially as it describes Mythos as strong enough to outperform nearly all human experts at finding and exploiting software vulnerabilities, which makes this less of a future concern and more of a current one.

"Mythos Preview has already found thousands of high-severity vulnerabilities, including some in every major operating system and web browser. Given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely. The fallout, for economies, public safety, and national security, could be severe. Project Glasswing is an urgent attempt to put these capabilities to work for defensive purposes."

That warning gives the project its purpose, and Anthropic's use of phrases like "urgent attempt" and "critical software" makes clear that the company sees this as a fast-moving problem that demands industry-wide action.

Apple's presence in Project Glasswing adds weight to the initiative because it places one of the world's biggest platform companies inside an early effort to use frontier AI for security testing at scale. Anthropic says Mythos has even found vulnerabilities that survived decades of human review and millions of automated tests, including one case involving the Linux kernel where chained flaws could have led to full control of a machine.

Anthropic also says Mythos goes beyond cybersecurity, with gains over Claude Opus 4.6 in reasoning, agentic search, computer use, and especially agentic coding, but the company does not plan to make Mythos Preview generally available. For now, the focus stays on controlled access, defensive use, and sharing what partners learn so the broader software industry can benefit before "Mythos-class models" become common.