Jamie Dimon has never been one to sugarcoat things. The JPMorgan Chase CEO is now directing his trademark bluntness at one of the most capable AI models ever built, warning that Anthropic’s Mythos could fundamentally reshape the threat landscape for banks, crypto exchanges, and every financial institution in between.
Dimon’s concern centers on Mythos’s demonstrated ability to identify and exploit software vulnerabilities. At a summit on July 15, he compared giving broad access to the model to handing out “ballistic missiles to individuals.”
What Mythos actually does, and why Dimon is worried
Mythos is part of Anthropic’s Claude model family, and it first turned heads during a preview in early April 2026 when it posted remarkable scores on cybersecurity benchmarks, proving exceptionally good at finding holes in software.
Rather than releasing Mythos to the public, Anthropic launched Project Glasswing, a restricted-access program that grants roughly 12 core partners and more than 40 additional organizations the ability to use the model. Anthropic provided $100 million in usage credits, earmarked specifically for vulnerability detection and remediation.
