NEW YORK — JPMorgan Chase CEO Jamie Dimon said cybersecurity threats tied to artificial intelligence increased tenfold after Anthropic released its Mythos model, warning that the technology exposed vulnerabilities that were previously unknown, Bloomberg reported.
Threats related to AI “went up 10-fold after Mythos,” Dimon said in an Oct. 6 interview with Bloomberg TV.
“AI created vulnerabilities that we didn’t know about, and we always worried about cyber before these things,” he said.
Bloomberg reported that Mythos took unauthorized actions after gaining internet access during safety testing earlier this year. Those actions prompted Anthropic to reconsider how advanced AI models could perform tasks in ways developers had not anticipated.
Anthropic and OpenAI have acknowledged that their models inadvertently breached multiple companies’ systems during testing, according to the reporting.
Working To Address Risks
Dimon described the potential for AI agents and advanced models to cause problems as a legitimate concern, while emphasizing the need to address those risks.
“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 told Bloomberg.
“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.”
Dimon also said data centers should be built in communities that support their construction and have sufficient electricity available.
“There are a lot of states that would take it that have access to power,” he said.
Pressure For Stronger Safeguards
Separately, the White House’s new AI safety task force faces calls to move beyond voluntary safeguards as concerns grow about autonomous systems acting outside their instructions.
President Donald Trump announced the task force over the weekend, naming national intelligence director Jay Clayton to lead it.
The administration has argued that restrictions could hamper development and allow China to gain a technological advantage. But reports of AI systems escaping testing environments, exploiting cybersecurity weaknesses and acting beyond their instructions are testing that approach.
Task force member Scott Kupor said companies would not receive a “free pass,” while cautioning against rules that could constrain U.S. innovation, according to The National News Desk.
The task force faces the challenge of maintaining U.S. leadership in advanced AI while demonstrating that federal policy can keep autonomous systems under human control, that report said.
