Bank of America hires 1,000 contractors to address flaws exposed by AI

Bank of America has hired around 1,000 external contractors to help fix software vulnerabilities exposed by advanced AI models, including Anthropic’s Mythos. CEO Brian Moynihan shared the information in a television interview.

Moynihan said the bank began work earlier in 2026. The effort covered internal code as well as software from vendors. The contractors, brought in during the second quarter of 2026, joined about 3,000 employees already working in this area.

The cost of remediation

Moynihan told analysts that the bank spent more than USD 200 million in the third quarter on software remediation and regulatory matters linked to anti-money laundering compliance. The compliance costs followed recent US Treasury penalties against the bank for failures in this area.

These are among the first public figures from a major US company on the staffing and spending involved in protecting systems from advanced AI models. Moynihan said the risk is real. Banks had to work out quickly how to respond to vulnerabilities identified by frontier models.

The CEO said the bank must stay ahead of these risks while recognising that the companies developing the software also sell services to the bank. He described the rise in spending as temporary. Once systems are patched, the need for further fixes should decrease, and remediation costs are expected to fall over time.

AI brings costs and savings

At the same time, AI is helping the bank reduce costs in other areas. Bank of America runs around 130 internal AI projects. Moynihan told analysts that savings from these initiatives are expected to rise from USD 300 million in 2026 to USD 700 million in 2027.

The bank’s workforce shrank by around 3,400 employees over the past year. The number of managers fell by roughly 5%, although it is not clear how much of the change is linked to AI.

The case shows two sides of AI adoption in banking. Models that can improve productivity can also find and exploit software vulnerabilities that have existed for years. For institutions with large legacy systems and many vendor relationships, this creates significant security costs. At Bank of America, remediation spending is currently running alongside the savings expected from AI projects.