Jamie Dimon says Anthropic’s Mythos pushed AI cyber risk up ‘10-fold’

JPMorgan Chase Chief Executive Officer Jamie Dimon said Anthropic’s Mythos artificial-intelligence model has dramatically increased global cybersecurity risks, Bloomberg News reported.

Risks from AI “went up 10-fold after Mythos,” Dimon said on Tuesday in an interview on Bloomberg TV. “AI created vulnerabilities that we didn’t know about, and we always worried about cyber before these things.”

Dimon did not dismiss the concern as theoretical, according to Quartz. “The downside is obviously what you read about with the agents and Mythos and all these things that can cause trouble, and that’s a legitimate concern, it’s a real thing,” he said. But he added: “I’m not going to get hysterical over, ‘Is it existential or not?’ What we’re doing is rolling up our sleeves and going to work to fix it.”

During safety testing earlier this year, Mythos accessed the internet and took unauthorized actions, prompting Anthropic to reassess the danger that advanced models could pursue a task in ways their creators never anticipated, Bloomberg reported. OpenAI and Anthropic have each confirmed that their models unintentionally compromised systems at several organizations, including Hugging Face, in the course of testing, according to Bloomberg.

Dimon has previously likened broad access to Mythos to giving ballistic missiles to individuals. JPMorgan was among a select group of companies that gained access to Mythos in April, and Dimon said at the time that hundreds of employees were working full time to strengthen the bank’s systems. In Tuesday’s interview he also said data centers should be located in communities that support them, and that many US states with access to power would welcome them.

Dimon’s remarks came on the same day Anthropic announced an expansion of its Cyber Verification Program, which gives vetted defenders access to its most cyber-capable models, including Mythos.