Anthropic Claude Mythos and Project Glasswing Rattle Banks and Big Tech in Urgent Cybersecurity Push
Anthropic's most powerful AI model to date — Claude Mythos — has triggered a rare emergency meeting between U.S. bank CEOs, Federal Reserve Chair Jerome Powell, and Treasury Secretary Scott Bessent. The extraordinary gathering this week reflects how seriously the financial sector is taking the cybersecurity risks tied to Mythos and the company's new initiative, Project Glasswing.
What Is Claude Mythos and Why Is It Not Public
Anthropic released Claude Mythos Preview to a limited group of tech companies, citing the potential damage a wider public release could cause. The system card for Mythos Preview states that its large increase in capabilities led Anthropic to decide against making it generally available.
Anthropic has privately warned top government officials that Mythos makes large-scale cyberattacks significantly more likely this year. The company acknowledges the same capabilities that can strengthen cyber defenses can also be weaponized by attackers.
Mythos Preview is a general-purpose model, yet in pre-release testing, Anthropic found its cybersecurity capabilities were surprisingly advanced compared with previous models. Logan Graham, who leads offensive cyber research at Anthropic, confirmed the model is advanced enough not only to identify undisclosed software vulnerabilities but also to weaponize them.
Project Glasswing: AI Fighting AI at Scale
Anthropic formed a coalition called Project Glasswing that includes Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. The companies use Mythos in preview mode to detect software bugs and fix them before hackers gain access to the technology.
As part of Project Glasswing, Anthropic is providing over 50 tech organizations access to Mythos Preview along with over $100 million in usage credits to find and fix vulnerabilities in foundational systems representing a large portion of the world's shared cyberattack surface.
In just the past few weeks, Claude Mythos Preview identified thousands of zero-day vulnerabilities — many of them critical — across every major operating system and every major web browser, along with a range of other important pieces of software.
Bank CEOs and the White House Emergency Meeting
Federal Reserve Chairman Jerome Powell and Treasury Secretary Scott Bessent met with major U.S. bank CEOs this week to discuss the possible cyber risks raised by Anthropic's Mythos model. The bank heads were already in Washington for a Financial Services Forum board meeting when a special gathering was called to discuss Mythos.
Bank of America, Citi, Goldman Sachs, Morgan Stanley, and Wells Fargo CEOs attended the Anthropic meeting. JPMorgan's Jamie Dimon was the only major banking CEO who could not attend.
TD Securities analyst Jaret Seiberg warned that if Claude Mythos helps bad actors find coding vulnerabilities faster than banks can fix them, it could destabilize a major bank and quickly become a systemic threat if it shatters confidence in the ability to store wealth and transact using financial institutions.
What Mythos Anthropic Means for Cybersecurity Stocks
JPMorgan reiterated overweight ratings on CrowdStrike Holdings and Palo Alto Networks following Anthropic's Project Glasswing announcement, setting a 12-month price target of $475 on CrowdStrike and $200 on Palo Alto Networks. JPMorgan analyst Brian Essex described both firms as essential layers in the defensive stack rather than competitive targets.
Cybersecurity shares surged after Anthropic unveiled Project Glasswing, reflecting market confidence that the initiative will drive sustained demand for advanced security solutions across enterprise and critical infrastructure sectors.
Anthropic's Claude Mythos and Project Glasswing represent a defining moment in AI development — one where the industry's most capable model is being deployed not for public use, but as a shield against the very threats it could unleash.
