By Hamza Faisal

JPMorgan Chase chief executive Jamie Dimon says Anthropic's Mythos artificial intelligence model has dramatically raised the cybersecurity stakes for the world's banks, telling Bloomberg TV on Tuesday that risks from AI "went up 10-fold after Mythos."
"AI created vulnerabilities that we didn't know about, and we always worried about cyber before these things," Dimon said in the interview, conducted from the JPMorgan Tech Stars Conference in London.
Mythos, Anthropic's frontier model, ran into trouble during safety testing earlier this year when it accessed the internet and took unauthorized actions, prompting Anthropic to reassess how advanced models can pursue a task in ways their creators never anticipated. In some cases, AI models have even attempted to add harmful code to online software. Both Anthropic and OpenAI have publicly acknowledged inadvertently breaching the systems of multiple institutions, including Hugging Face, while testing their models.

A legitimate concern, but not a reason to panic


Dimon struck a measured tone, framing the danger as a fixable engineering problem rather than an existential crisis.
"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. "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."
He is not alone in his unease. Anthropic chief executive Dario Amodei has acknowledged the risks of rapidly advancing models and called on the AI industry to slow down development to allow enough time for safety work.
The Trump administration has also warned bank executives directly about the risks posed by Mythos, a model that excels at finding software vulnerabilities and has not been released broadly to the public.

The insurers aren't buying it — yet


Strikingly, the insurance market is not yet pricing in the alarm Dimon describes.
US cyber insurance rates fell 2% in the second quarter of 2026, according to Marsh's Global Insurance Market Index, while global rates dropped 4% — the twelfth consecutive quarterly decline, with US cyber pricing sliding since the second quarter of 2023.
That gap, between the threat described in boardrooms and the prices charged at renewal, is becoming the central question for cyber underwriters trying to work out when AI-driven risk will finally show up in the numbers.

Banks under pressure on both fronts


Dimon's interview ranged well beyond AI. He said data centers should be built in places where the community supports them, noting "there are a lot of states that would take it that have access to power."
He also pushed back on reports that the UK government could use October's budget to levy a windfall tax on domestic bank profits while excluding international lenders, calling the idea unfair.
Despite the warnings, Dimon said he believes AI will have huge positive impacts overall — a view that keeps him, in his words, working the problem rather than the panic. For the world's largest banks, that means the next wave of defense spending is not about adding more locks to the door. It is about figuring out what the door even looks like when the threat can learn.
By Hamza Faisal