Bank of America brought in about 1,000 outside contractors in the second quarter to repair software weaknesses exposed by AI models, CEO Brian Moynihan said on 14 October. He named Anthropic’s Mythos as one of the models involved. Moynihan disclosed the figure in a CNBC interview after the bank reported third-quarter results and repeated it on the earnings call; TNW said it had not independently verified the numbers. About 3,000 Bank of America employees were already assigned to the work before the contractors joined.

Moynihan said the weaknesses were in code used by the bank and its outside vendors. He told analysts that the quarter included “a couple hundred million dollars plus” in remediation costs, covering both the Mythos-related work and repairs to the bank’s anti-money-laundering controls. He expects the Mythos share of those costs to decline now that the extra workers are in place. The 1,000 contractors were consultants and were not counted as bank employees.

The compliance spending followed Treasury penalties of more than $1.21 billion announced on 13 October. Federal regulators said the bank failed to report suspicious transactions worth $5 billion and had not completed steps ordered in 2024. Bank of America’s quarterly revenue rose about 3% to nearly $30 billion, but adjusted earnings of $1.08 a share fell short of analyst forecasts, and the stock closed almost 5% lower.

Anthropic made Mythos available in April to a small group of companies, banks and government agencies through Project Glasswing, intended to help them find and patch flaws before attackers did. A month later, US regulators paused some cyber examinations to give the largest banks time to address the flaws. Moynihan had described the model as a major change in banks’ workload in July. In May, JPMorgan Chase CEO Jamie Dimon said his bank had hundreds of people working full-time on the task. US and European regulators are questioning banks about AI agents and cyber risks, and the European Central Bank wants plans against AI-enabled threats by the end of October.

Bank of America brought in about 1,000 outside contractors in the second quarter to repair software weaknesses exposed by AI models, CEO Brian Moynihan said on 14 October. He named Anthropic’s Mythos as one of the models involved. Moynihan disclosed the figure in a CNBC interview after the bank reported third-quarter results and repeated it on the earnings call; TNW said it had not independently verified the numbers. About 3,000 Bank of America employees were already assigned to the work before the contractors joined.

Moynihan said the weaknesses were in code used by the bank and its outside vendors. He told analysts that the quarter included “a couple hundred million dollars plus” in remediation costs, covering both the Mythos-related work and repairs to the bank’s anti-money-laundering controls. He expects the Mythos share of those costs to decline now that the extra workers are in place. The 1,000 contractors were consultants and were not counted as bank employees.

The compliance spending followed Treasury penalties of more than $1.21 billion announced on 13 October. Federal regulators said the bank failed to report suspicious transactions worth $5 billion and had not completed steps ordered in 2024. Bank of America’s quarterly revenue rose about 3% to nearly $30 billion, but adjusted earnings of $1.08 a share fell short of analyst forecasts, and the stock closed almost 5% lower.

Anthropic made Mythos available in April to a small group of companies, banks and government agencies through Project Glasswing, intended to help them find and patch flaws before attackers did. A month later, US regulators paused some cyber examinations to give the largest banks time to address the flaws. Moynihan had described the model as a major change in banks’ workload in July. In May, JPMorgan Chase CEO Jamie Dimon said his bank had hundreds of people working full-time on the task. US and European regulators are questioning banks about AI agents and cyber risks, and the European Central Bank wants plans against AI-enabled threats by the end of October.