JPMorgan Chase chief executive Jamie Dimon has said artificial intelligence is currently making cybersecurity threats more severe, even as it may eventually help strengthen defences. Speaking on the bank’s earnings call, Dimon said JPMorgan is testing Anthropic’s AI model, Mythos, which he said has already shown an ability to uncover thousands of vulnerabilities in corporate software.
Risks and Promise of Anthropic’s Mythos
Dimon said artificial intelligence has made cybersecurity harder by creating additional vulnerabilities, while also holding out the possibility of stronger protections in the future. Referring to Mythos, he said the tool shows that many more vulnerabilities need to be fixed, reflecting both the risks and the potential benefits of deploying advanced AI in cybersecurity.
The model, which Anthropic has released only to a select group of companies, has already raised concerns across Wall Street and government agencies over the possibility that it could be weaponised by hackers.
FCRF Returns With CDPO, Its Premier Data Protection Certification for Privacy Professionals
Treasury Meeting Signals Wider Concern
Dimon’s remarks came days after US Treasury Secretary Scott Bessent met bank chief executives in Washington to discuss risks related to Mythos. The discussion underscored the extent to which concern over advanced AI tools has spread beyond private companies into the highest levels of financial and government oversight.
Dimon said the risks were not confined to any one institution because of the interconnected nature of the financial system. He noted that banks are tied to exchanges and other entities that add further layers of risk.
JPMorgan Stresses Basic Cyber Discipline
Dimon said JPMorgan has long invested heavily in cybersecurity and maintains constant coordination with government agencies. He said the bank employs top experts and treats cyber defence as a continuous, full-time responsibility.
JPMorgan chief financial officer Jeremy Barnum echoed those concerns, saying the industry has long recognised that AI can both expose vulnerabilities and be used in attack mode by bad actors. Dimon also stressed that traditional cybersecurity practices remain essential, saying strong protection of data, networks, routers, hardware and passcodes continues to play a central role in reducing risk.
