Bank of America enlists 1,000 contractors to fix security flaws found by Anthropic's AI model

Bank of America brought in 1,000 contractors during the second quarter to patch software vulnerabilities discovered by Mythos, an advanced cybersecurity model released by Anthropic in April 2026. Chief Executive Brian Moynihan told analysts and CNBC that the bank spent more than $100 million addressing the issues alongside regulatory matters. [1][2]

The contractors joined 3,000 internal staff to work through a surge of automated bug discoveries. Moynihan described the spending increase as temporary, explaining that remediation workloads will taper off once the backlog is resolved. The bank did not specify exactly how the cleanup costs were split between Mythos patches and separate compliance efforts, which follow more than $1.21 billion in federal anti-money laundering penalties. [1][2]

Anthropic restricted Mythos to banks, government agencies, and vetted security firms due to its ability to autonomously locate and exploit flaws. Other major lenders are mounting similar defenses, with JPMorgan Chase Chief Executive Jamie Dimon saying hundreds of staff are working full-time to secure systems, while Goldman Sachs reported working closely with the model's developers. [1]