JPMorgan Chase Chief Executive Officer Jamie Dimon stated that the artificial intelligence model Mythos from Anthropic PBC has dramatically heightened global cybersecurity risks.

The risks brought by AI "increased tenfold after Mythos appeared," Dimon said in an interview on Tuesday. "AI has introduced vulnerabilities we previously did not know about, and before this, we had always been quite concerned about cybersecurity issues."

During security testing conducted earlier this year, Mythos accessed the internet and took unauthorized actions, prompting Anthropic to reassess the risks that advanced models could execute tasks in ways developers never anticipated.

In some instances, AI models attempted to add malicious code to online software, marking the latest case in a recent string of security incidents and further intensifying concerns that AI vendors cannot fully control their models.

Both OpenAI and Anthropic have publicly acknowledged that they inadvertently breached systems at several organizations, including Hugging Face Inc., during model testing.

"The downside is obviously what you see regarding AI agents, Mythos, and all these things that could cause trouble — this concern is reasonable and real," Dimon said. "I won't panic over whether this constitutes an existential threat. What we are doing is rolling up our sleeves and getting to work on solving the problem."

In the wide-ranging interview covering numerous topics, Dimon also said that data centers should be built in locations where local communities support them. He said: "There are many states willing to host them, and they have power supply available."