Bank of America's losses due to the implementation

Bank of America implemented the AI model Claude to optimize expenses, which led to unforeseen consequences. The system identified numerous vulnerabilities in processes requiring urgent remediation.

To address these issues, the financial institution had to quickly engage approximately one thousand additional contractors. This occurred against the backdrop of an existing team of three thousand full-time specialists in the relevant field.

The bank's financial losses from fixing the discovered problems amounted to hundreds of millions of dollars. The incident highlights the risks of implementing generative artificial intelligence in corporate infrastructure without adequate prior security checks.