Dimon calls the rollout of Anthropic's powerful new AI a 'very heightened risk' as crypto firms rush to use it for security testing
Jamie Dimon has a new thing keeping him up at night, and for once it’s not interest rates. The JPMorgan Chase CEO is sounding alarms about Anthropic’s Mythos AI model, warning that broad access to the system poses serious cybersecurity risks across the financial sector and beyond.
Dimon first flagged the issue during JPMorgan’s Q1 earnings call on April 14, 2026, stating that Mythos “does create additional vulnerabilities.” He doubled down at a May 2026 event alongside Anthropic CEO Dario Amodei, calling the situation a “very heightened risk” and describing the current moment as a “transitory period” while organizations scramble to understand what they’re dealing with.
Thousands of vulnerabilities, months to fix #
During controlled testing, the model reportedly identified thousands of software vulnerabilities, many of which remain unpatched. According to Amodei, addressing those unpatched vulnerabilities could take 6 to 12 months.
Dimon commended Anthropic for its cautious rollout approach, giving organizations time to evaluate the risks before broader deployment. The JPMorgan chief emphasized that cyber risk remains the bank’s primary concern.
Crypto firms want in #
The implications aren’t confined to traditional banking. Cryptocurrency firms are actively pursuing access to Mythos, with notable exchanges like Coinbase and Binance reportedly engaging with Anthropic for enhanced security testing.
The conversation is shifting from the usual suspects, smart contract bugs and rug pulls, toward more systemic infrastructure concerns, including key management systems and oracle reliability.
Wall Street’s AI anxiety goes mainstream #
Dimon isn’t alone in raising these concerns. Goldman Sachs CEO David Solomon referenced Mythos in April 2026, highlighting what he sees as Wall Street’s dual challenge: adopting AI aggressively enough to stay competitive while managing the cyber risks that come with it.
The 6 to 12-month remediation timeline Amodei mentioned creates a window of elevated risk. Any leaks or unauthorized access to Mythos findings during that period could trigger market-moving security incidents.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our