TL;DR
HSBC, Barclays, Lloyds, NatWest and Nationwide have moved to keep AI agents out of customer accounts, according to a Telegraph report on 8 November, in some cases by banning transactions an agent makes without the customer’s direct involvement. The paper understands the bans to be precautionary: there are no signs yet of agents getting into UK accounts.
Why the banks are acting
Two worries sit behind the move. The first is control. Banks fear a personal finance bot could mismanage someone’s money with no settled answer on who pays when it does. “We need to be very careful that those tools are safe for customers and safe for us before we allow them in,” Standard Chartered’s chief financial officer, Manus Costello, said. Lloyds’ finance chief, Will Chalmers, said the bank had seen no sign of agents in accounts, but “we do guard against that type of risk”.
The second is money. The Telegraph cites a McKinsey warning that savers could move up to £700bn out of low-paying accounts on AI advice, and that shifting just 5% to 10% of balances could cut deposit profits by a fifth or more. Agents acting at once could, in the worst case, pull funds from a bank at machine speed.
Following the US
The UK moves mirror earlier blocks in North America. JP Morgan, Citigroup, Bank of America and Wells Fargo have all stopped agents accessing accounts, according to the Telegraph. Jamie Dimon, JP Morgan’s boss, said on its third-quarter earnings call: “You don’t want rogue agents, and there could be thousands of them.” Charles Schwab’s chief executive, Rick Wurster, said the broker stops 10 to 20 clients each week from letting agents into their accounts.
UK rules already slow agents down, because strict authentication means most payments need sign-off in a bank’s app, usually by fingerprint or face. Rules for agents in finance are being drawn up by the FCA, with industry and fellow regulators involved.
Similar consumer agents are expected in Britain. Meta’s Muse assistant, which can shop and pay bills for users, is expected to launch in the UK within weeks, and a government clause in cyber legislation would give ministers powers over AI agent loss-of-control incidents.
What it means
In our view, the line the banks are drawing is between reading and acting. The Telegraph notes that new personal finance products from Anthropic and OpenAI only read account data, and Dealroom, summarising the report, says read-only access via approved open banking services is not necessarily off the table. Startups and fintechs building agents for UK customers should assume that taking action on an account will need an approved route, and a clear liability model, before banks open the door.