JPMorgan blocks external AI agents, warns of rogue-agent risks to customers

During its earnings call, JPMorgan Chase CEO Jamie Dimon warned that personal AI agents differ from chatbots by taking actions on users' accounts, raising the risk of rogue behavior, liability transfer, and data misuse. He said the bank will keep external agents out of its systems until robust safeguards exist. CFO Jeremy Barnum added that JPMorgan blocks credential-sharing, screen-scraping, and is restricting internal use of tools like Anthropic’s Claude.

While the firm is building its own agentic solutions, it is currently keeping a tight gate. These remarks arrived amid concerns that AI assistants such as Meta’s Muse could trigger rapid fund movements, a scenario highlighted by Apollo economist Torsten Sløk, and have already nudged bank shares lower. Barnum emphasized the bank’s focus on retaining primary share rather than chasing high-yield accounts, citing its brick-and-mortar footprint as a defense against fintech competition.