AI Has Increased Global Cyber Risks Tenfold, Says JPMorgan CEO. How Can You Profit From This?

Anthropic’s Mythos AI model has dramatically increased global cyber risks, JPMorgan Chase CEO Jamie Dimon said Tuesday in an interview with Bloomberg TV. “AI-related risks have increased tenfold since the emergence of [Anthropic’s AI model] Mythos,” Dimon said. “AI has created vulnerabilities we didn’t even know existed, even though cyber risks had been a concern for us before.”

AI labs are increasingly reporting instances of AI models exhibiting behavior that is beyond human control: for example, once they achieve their goals, agents gain unauthorized access to databases or hack websites, as was the case with the Hugging Face project. The industry is concerned that AI developers are unable to fully control the agents they create. Both OpenAI and Anthropic have publicly acknowledged that their agents or models unintentionally hacked into the systems of several organizations during testing.

Damon emphasized that he does not consider the situation a cause for panic. “The flip side is obvious—it’s what you’re reading about, the agents, the Mythos, and everything else that could cause trouble, and that’s a real, legitimate concern,” he said. “I’m not going to get hysterical about whether or not this is an existential threat. We’re just working on fixing it.”

The discussion about the risks associated with Mythos comes as Anthropic prepares for a potential initial public offering, which, according to Bloomberg, could take place as early as October—ahead of the planned IPOs of OpenAI and the Chinese AI lab DeepSeek.

JPMorgan analyst Brian Essex believes that the stocks of five cybersecurity service providers could benefit from the spread of AI. These companies are CrowdStrike, Palo Alto Networks, Okta, Zscaler, and Varonis Systems.

“This quarter, we saw evidence that the accelerating adoption of AI is expanding the attack surface, reaffirming the critical importance of investing in next-generation security systems and the fact that companies cannot simply cut back on these expenses, — Essex wrote in a note cited by Barron’s. — “We expect this trend to continue regardless of attempts by some industry players to ‘slow down’ the development of advanced AI models.”

According to the analyst’s estimate, AI could lead to additional cybersecurity spending of more than $430 billion over the next three years. He estimates that security currently accounts for more than 10% of total IT budgets, so the boom in AI investment could bring in billions of dollars in additional revenue for companies specializing in AI security.

CrowdStrike and Palo Alto—the two largest publicly traded companies specializing directly in cybersecurity—are among Essex’s favorites thanks to the wide range of solutions they offer, according to Barron’s.

According to the analyst, Okta should benefit from the growing role of user identity and access management software in a world where AI is becoming ubiquitous. Zscaler, for its part, may gain traction thanks to its cloud-based network security solutions.

Varonis is significantly smaller than its competitors and specializes in data protection—identifying and addressing instances where sensitive information is not adequately protected. Such solutions may become particularly in demand as AI agents gain increasingly broad access to corporate data.

“The physical expansion of AI infrastructure directly drives growth in demand for real-time traffic monitoring and data protection,” Essex wrote. “Given how long it typically takes from the start of negotiations to the closing of a deal, we expect a noticeable impact on companies’ revenue to begin in the fourth quarter, and this effect will intensify in 2027,” the analyst said.

This article was AI-translated and verified by a human editor