Brian Moynihan, chairman and CEO of Bank of America, said it was encouraging that AI companies were treating safety as a genuine priority rather than a talking point. His remarks come as Anthropic, OpenAI and Google leaders discuss creating an industry safety body, with some participants proposing a deliberate slowdown in development so safety protocols can catch up. Bank of America has approved more than 300 AI use cases, including 114 for generative AI, and employees generate more than 400,000 AI prompts daily. The bank operates 270 AI and machine-learning models, which it says have delivered measurable productivity and security gains, including a 55% reduction in fraud losses and a 20% increase in developer productivity. Its AI-powered virtual assistant, Erica, connects with 110 internal systems and handles about 700 types of customer questions. Moynihan supports a human-in-the-loop approach, keeping people involved in high-stakes decisions while AI performs computational work. He also highlighted security concerns during a Bloomberg Television interview in July, citing Anthropic’s Mythos as an example of rapidly advancing AI that requires banks to continually update their defenses.