Anthropic Mythos reveals ‘more vulnerabilities’ to cyberattacks

JPMorgan Chase While artificial intelligence tools can ultimately help companies defend themselves from cyberattacks, they first make them more vulnerable, the company’s CEO, Jamie Dimon, said Tuesday.

Dimon said JPMorgan was testing Anthropic’s latest model — a preview of Mythos that the AI ​​company announced last week — as part of its broader efforts to reap the benefits of AI while protecting against bad actors using the same technology.

“AI has made it worse, made it more difficult,” Dimon told analysts on the bank’s earnings call Tuesday morning. “It creates additional weaknesses and, perhaps in the future, better ways to strengthen yourself as well.”

When a reporter asked him about Mythos, Damon seemed to be referring to Anthropic’s warning that the model has already discovered thousands of vulnerabilities in corporate software.

“I think you read exactly what it is,” Damon said. “It shows that there are a lot of vulnerabilities that need to be fixed.”

These statements reveal how artificial intelligence, a technology once welcomed by companies as a boon to productivity, has turned into a serious threat by giving bad actors new ways to penetrate technology systems. Last week, Treasury Secretary Scott Besent summoned bank chief executives to a meeting to discuss the risks posed by Mythos.

JPMorgan, the world’s largest bank by market value, has for years invested heavily to stay ahead of threats, with dedicated teams and ongoing coordination with government agencies, Dimon said.

“We spend a lot of money. We have top experts. We are in constant contact with the government,” he said. “It’s a full-time job, and we do it all the time.”

However, the CEO warned that the risks extend beyond any single institution, given the interconnected nature of the financial system.

“This does not mean that everything banks rely on is well protected,” Dimon said. “Banks…are tied to exchanges and all these other things that create other layers of risk.”

The industry has long recognized that AI cuts both ways in cybersecurity, said Jeremy Barnum, chief financial officer of JPMorgan.

“These tools can make vulnerabilities easier to find, but they are also more likely to be deployed by bad actors in attack mode,” Barnum said on the earnings call. Recent developments by Anthropic and others have simply intensified an existing trend, he said.

Dimon also said that while advanced AI tools are important, legacy cybersecurity practices remain essential.

“A big part of it is hygiene… How do you protect your data? How do you protect your networks, your routers, your devices, changing your passcode?” He said. “Doing all of these things right greatly reduces the risk.”

Goldman Sachs CEO David Solomon said Monday during an earnings call that his bank was testing Mythos, though he declined to comment further.