Bank of America hired contractors to protect its software from powerful rogue AI agents

• The US's second-largest bank is protecting itself against powerful AI models.
• Bank of America hired 1,000 consultants to shore up cracks in its software that could be exploited.
• Increasingly capable and rogue AI agents have become a large concern in recent months.

Bank of America is shoring up its defenses against powerful AI models.

CEO Brian Moynihan told analysts in a Wednesday earnings call that the bank had hired 1,000 extra staff — external consultants — for what he called "Mythos remediation."

He expanded on that in an interview with CNBC on Wednesday. He said the development of powerful AI models, such as Anthropic's Mythos, had exposed certain software vulnerabilities within the bank's code, which is used by both internal staff and external vendors.

Mythos, a restricted AI model known for its ability to find and exploit software weaknesses, has been released in stages over the past year with guardrails. Moynihan told CNBC that when frontier models started coming out, BofA executives sat down and brainstormed how to quickly shore up vulnerabilities.

"We put on 1,000 contractors to help us do the patching work in the second quarter, and those people remain here today, helping us, on top of the 3,000 people we have in that area," Moynihan told CNBC.

The executive did not say exactly how much the initiative cost. However, he said on the earnings call that BofA had spent a "couple hundred million plus" to fix the software problems and address anti-money-laundering compliance issues that led to a $1.2 billion Treasury penalty.

BofA's investment in beefing up its software security comes as AI models get increasingly capable and autonomous. AI agents by labs like OpenAI, Anthropic, Google, and Meta have hacked into the software and website of several organizations, including AI firm Hugging Face and Australia's government website. Anthropic's agents even submitted a false murder tip to Philadelphia police.

JPMorgan Chase CEO Jamie Dimon has previously spoken out against Mythos, telling Bloomberg earlier this month that cybersecurity risks "went up 10-fold" after the model's release.

"AI created vulnerabilities that we didn't know about, and we always worried about cyber before these things," he added.