Bank of America CEO: 1,000 Contractors Fixing AI-Exposed Flaws
Brian Moynihan, Bank of America's chair and CEO, revealed that approximately 1,000 outside contractors were hired in Q2 to address software vulnerabilities exposed by AI models, specifically mentioning Anthropic's Mythos. This move follows $1.21 billion in Treasury penalties for failing to report suspicious transactions and complete required steps.
Key Points:
- Contractor Hiring: About 1,000 contractors were brought in, building upon the existing effort of approximately 3,000 internal staff.
- AI Model Identified: Moynihan attributed the flaws to Anthropic’s Mythos AI model.
- Remediation Costs: He estimated quarterly remediation costs at "a couple hundred million dollars plus", including work on AI vulnerabilities and anti-money-laundering controls.
- Consultant Status: The contractors are classified as consultants rather than headcount due to their specific roles.
This development highlights the ongoing challenges financial institutions face in addressing AI-driven cybersecurity risks.