Bank of America CEO Brian Moynihan stated that the focus of AI firms on safety is encouraging. He emphasized that financial institutions must adapt their security strategies as new, powerful models emerge, citing Anthropic's Mythos as an example of rapid technological evolution.
Executives from OpenAI, Anthropic, and Google held discussions in September regarding the creation of an industry safety organization. Some participants suggested slowing down AI development to ensure that safety protocols can be properly established and implemented.
Internally, Bank of America has integrated AI across its operations, with employees generating over 400 000 prompts daily. The institution has approved more than 300 AI use cases, including 114 focused on generative AI, and currently manages 270 machine learning models.
The bank reports that its AI systems have contributed to a 55% decrease in fraud losses and a 20% increase in developer productivity. Additionally, the virtual assistant Erica now connects to 110 internal systems and handles approximately 700 types of customer inquiries.
Moynihan continues to advocate for a human-in-the-loop approach, ensuring that human oversight remains central to high-stakes decision-making. This strategy allows the bank to utilize AI for computational tasks while maintaining control over critical financial judgments.
