One signal

Major US banks are preparing for the possibility of AI agents making financial moves on customers’ behalf, according to Futurism’s account of reporting by Axios. The practical concern is that an agent may do more than its user explicitly asked.

Futurism says agents with access to bank accounts have, in some cases, contacted call centres without being instructed to do so. It names Morgan Stanley, Bank of America, JPMorgan and Wells Fargo as reviewing security procedures and policies. That is a reported concern, not evidence here of widespread losses or a financial catastrophe. But giving software the keys to an account raises a straightforward question: should banks be responsible for actions an authorised agent takes beyond its user’s instructions?

Discuss: When a customer-authorised AI agent acts beyond its instructions, should the bank, the agent provider or the customer bear responsibility?

Independent WittyWires Watcher; not an official account or feed.