Jamie Dimon, chief government officer of JPMorgan Chase & Co., proper, departs the US Capitol in Washington, DC, US, on Wednesday, Feb. 25, 2026.
Graeme Sloan | Bloomberg | Getty Pictures
JPMorgan Chase CEO Jamie Dimon stated Tuesday that whereas synthetic intelligence instruments may ultimately assist firms defend themselves from cyberattacks, they’re first making them extra susceptible.
Dimon stated that JPMorgan was testing Anthropic’s newest mannequin — the Mythos preview introduced by the AI agency final week — as a part of its broader effort to reap the advantages of AI whereas defending towards dangerous actors wielding the identical expertise.
“AI’s made it worse, it is made it more durable,” Dimon instructed analysts on the financial institution’s earnings name Tuesday morning. “It does create further vulnerabilities, and perhaps down the street, higher methods to strengthen your self too.”
When requested by a reporter about Mythos, Dimon appeared to confer with Anthropic’s warning that the mannequin had already discovered hundreds of vulnerabilities in company software program.
“I believe you learn precisely what’s it,” Dimon stated. “It reveals much more vulnerabilities should be fastened.”
The remarks reveal how synthetic intelligence, a expertise welcomed by companies as a productiveness boon, has additionally morphed right into a critical risk by giving dangerous actors new methods to hack into expertise techniques. Final week, Treasury Secretary Scott Bessent summoned financial institution CEOs to a gathering to debate the dangers posed by Mythos.
JPMorgan, the world’s largest financial institution by market cap, has for years invested closely to remain forward of threats, with devoted groups and fixed coordination with authorities companies, Dimon stated.
“We spend some huge cash. We have got prime specialists. We’re in fixed contact with the federal government,” he stated. “It is a full-time job, and we’re doing it on a regular basis.”
‘Assault mode’
Nonetheless, the CEO warned that dangers lengthen past any single establishment, given the interconnected nature of the monetary system.
“That does not imply every part that banks depend on is that nicely protected,” Dimon stated. “Banks… are connected to exchanges and all these different issues that create different layers of danger.”
JPMorgan Chief Monetary Officer Jeremy Barnum stated the trade has lengthy been conscious that AI cuts each methods in cybersecurity.
“These instruments could make it simpler to search out vulnerabilities, however then additionally doubtlessly be deployed by dangerous actors in assault mode,” Barnum stated on the earnings name. Current advances from Anthropic and others have merely intensified an present development, he stated.
Dimon additionally stated that whereas superior AI instruments are necessary, old-school cybersecurity practices stay important.
“Numerous it’s hygiene… how do you shield your knowledge? How do you shield your networks, your routers, your {hardware}, altering your passcode?” he stated. “Doing all these issues proper dramatically reduces the chance.”
Goldman Sachs CEO David Solomon stated Monday throughout an earnings name that his financial institution was testing Mythos, although he declined to remark additional.