Financial institution of America is shoring up its defenses towards highly effective AI fashions.
CEO Brian Moynihan instructed analysts in a Wednesday earnings name that the financial institution had employed 1,000 additional workers — exterior consultants — for what he known as “Mythos remediation.”
He expanded on that in an interview with CNBC on Wednesday. He stated the event of highly effective AI fashions, reminiscent of Anthropic’s Mythos, had uncovered sure software program vulnerabilities inside the financial institution’s code, which is utilized by each inside workers and exterior distributors.
Mythos, a restricted AI mannequin recognized for its capability to search out and exploit software program weaknesses, has been launched in levels over the previous yr with guardrails. Moynihan instructed CNBC that when frontier fashions began popping out, BofA executives sat down and brainstormed methods to rapidly shore up vulnerabilities.
“We placed on 1,000 contractors to assist us do the patching work within the second quarter, and people folks stay right here at the moment, serving to us, on high of the three,000 folks we now have in that space,” Moynihan instructed CNBC.
The manager didn’t say precisely how a lot the initiative price. Nevertheless, he stated on the earnings name that BofA had spent a “couple hundred million plus” to repair the software program issues and deal with anti-money-laundering compliance points that led to a $1.2 billion Treasury penalty.
BofA’s funding in beefing up its software program safety comes as AI fashions get more and more succesful and autonomous. AI brokers by labs like OpenAI, Anthropic, Google, and Meta have hacked into the software program and web site of a number of organizations, together with AI agency Hugging Face and Australia’s government website. Anthropic’s brokers even submitted a false homicide tip to Philadelphia police.
JPMorgan Chase CEO Jamie Dimon has beforehand spoken out towards Mythos, telling Bloomberg earlier this month that cybersecurity dangers “went up 10-fold” after the mannequin’s launch.
“AI created vulnerabilities that we did not learn about, and we at all times fearful about cyber earlier than this stuff,” he added.
