Bank of America took swift action this year to address software vulnerabilities revealed by Anthropic’s advanced AI model, Mythos, bringing in 1,000 external contractors to support its 3,000 internal staff dedicated to patching both proprietary and vendor software. CEO Brian Moynihan emphasized the real risk these AI-discovered flaws posed and explained the bank’s urgent response to quickly mitigate the threat. The contractors hired in the second quarter remain engaged in these efforts to strengthen the bank’s cybersecurity defenses.

The remediation process, combined with ongoing anti-money laundering compliance, significantly impacted the bank's third-quarter expenses, costing several hundred million dollars. Moynihan acknowledged these added costs as temporary but necessary to protect the integrity of the bank’s software infrastructure. Despite a dip in Bank of America’s stock price following the earnings announcement, the CEO reassured investors that the intense patching phase would ease over time as fewer vulnerabilities required attention.

While addressing these defensive challenges, Bank of America is simultaneously leveraging AI to enhance operational efficiency. The bank currently runs around 130 internal AI projects, with projected savings rising from $300 million this year to an anticipated $700 million in 2027. Moynihan highlighted that AI is playing a dual role—posing initial costs due to new risks but also enabling significant financial benefits through automation and improved processes in the longer term.

In terms of workforce changes, the bank cut approximately 3,400 employees over the last year, reducing managerial positions by around 5%, though Moynihan did not specify the extent to which AI contributed to this reduction. The bank remains focused on balancing the integration of AI-driven innovations with maintaining robust security measures to safeguard its systems and comply with regulatory demands.