JPMorgan Chase CEO Jamie Dimon acknowledged that while artificial intelligence promises to strengthen defenses in the long term, it is currently making cybersecurity threats even more serious. Dimon recently revealed on the bank’s earnings call that JPMorgan is testing Anthropic’s new AI model, Mythos, as reported by CNBC. The model has already demonstrated the ability to discover thousands of vulnerabilities in corporate software. “AI has made things worse and more difficult. It certainly creates new vulnerabilities and maybe even better ways to strengthen ourselves in the future,” Dimon told analysts.When asked about Mythos, Dimon pointed to Anthropic’s own caveats about the tool’s capabilities. “This shows us that many more vulnerabilities need to be fixed,” he said, acknowledging both the risks and potential benefits of implementing advanced AI in cybersecurity.
US Treasury also warns against Anthropic myths
Dimon’s comments came days after Treasury Secretary Scott Bessent met with bank CEOs in Washington to discuss risks related to Mythos. The model, which Anthropic is making available only to a select group of companies, has already raised concerns on Wall Street and across government agencies about its potential to be weaponized by hackers.
JP Morgan’s Cyber Security Promotion
As the world’s largest bank by market capitalization, JPMorgan has invested heavily in cybersecurity over the years. Mr. Dimon said the bank employs top experts and maintains constant collaboration with government agencies. “We spend a lot of money. We have top experts. We are in constant contact with the government. “This is a full-time job and we do it all the time,” he said.Still, Dimon cautioned that the interconnected nature of the financial system means risks extend beyond a single institution. “Banks are obsessed with exchanges and everything else, and that creates another layer of risk,” he said.JPMorgan Chief Financial Officer Jeremy Burnham echoed Dimon’s concerns, saying the industry has long known that AI can expose vulnerabilities and be exploited in “attack mode” by bad actors. “While these tools make it easier to discover vulnerabilities, they can also be deployed by malicious parties,” Burnham said.Dimon emphasized that while advanced AI tools are important, traditional cybersecurity practices remain essential. “A lot of it is hygiene. How do you protect your data? How do you protect your network, routers, hardware, change passcodes? Doing all of these things right dramatically reduces risk,” he said.
