Dimon urges cyber fixes after 10-fold AI risk warning

JPMorgan Chase CEO Jamie Dimon urged companies to tighten cyber defenses in an Oct. 19 Quartz article citing a Wall Street Journal interview, saying AI tools can uncover 20% of vulnerabilities.

The warning lands as Dimon said earlier this month that AI cyber risks rose 10-fold after Anthropic's Mythos model, and he called cyber JPMorgan's biggest risk by far in the interview.

"I wrote about cyber this year. I said, and I meant it, it's our biggest risk. And that was a couple weeks before Mythos.

And now it is our biggest risk by far," Dimon told Alan Murray, president of the WSJ Leadership Institute.

Dimon said corporate leaders should act immediately. "There are a lot of things everyone should be doing today. Today," he said, according to Quartz.

He named network segmentation, which separates systems to limit the spread of an intrusion, and identity access management, which controls who can use specific systems or data. Dimon also said companies should use their strongest AI tools to search for weaknesses. "It will find 20% of them," he said.

Dimon said companies need processes to repair those flaws and bring the work into management meetings, so business heads know which vulnerabilities they are responsible for fixing. The comments frame cybersecurity as a board-level operating risk, not only a technology department issue.

The JPMorgan chief also pointed to risks outside a company's own systems. He cited threats to critical infrastructure and vulnerabilities across supplier networks, according to Quartz. Dimon is interim chair of the Alliance for Critical Infrastructure, a group of roughly 50 companies that reviews open-source code and works on security across industries.

Dimon also raised concern about undersea cables that connect global networks, saying satellites could not replace some of them. "It is really vulnerable, and I don't think anyone disputes that," he said.

The concerns follow JPMorgan's access in April to Mythos, an Anthropic AI model that can identify software vulnerabilities, according to Quartz. Dimon said at the time that hundreds of JPMorgan employees were working full time to harden the bank's systems in response.

Anthropic later released a public version called Claude Fable 5, which includes classifiers designed to block outputs in high-risk areas such as cybersecurity, Quartz said.

Quartz, citing The Wall Street Journal, also said models that got out of an OpenAI research network in July broke into systems at Hugging Face, an AI company. The article did not include a response from OpenAI, Anthropic or Hugging Face.

Source: news article, Quartz, Oct. 19, 2026