UK banks block AI agents from customer accounts as 'agentic bank run' fears build
HSBC, Barclays, Lloyds, NatWest and Nationwide have moved to block or restrict third-party AI agents from reaching customer accounts, especially any that can move money without a human in the loop. The banks cite security and reliability. Behind it sits a bigger worry: agents that shop customers into better deals and drain cheap deposits.
- Original source: ct.com.
- The source published it on Nov 16, 2026 at 2:30 PM.
- No second outlet has reported it yet. Treat it as a single-source claim.
Some of Britain's biggest lenders are shutting AI agents out of the vault. HSBC, Barclays, Lloyds, NatWest and Nationwide have each moved to block or restrict third-party AI agents from accessing customer accounts — with particular focus on agents able to transact without the customer's involvement. There is no evidence yet of AI agents accessing UK bank accounts. The concern is what happens if an automated system mismanages someone's money, and who carries the liability when it does.
Apollo chief economist Torsten Slok put a name on the systemic version of that fear in September: an 'agentic bank run.' If swarms of agents shift household cash out of low-yield checking accounts en masse, banks lose the cheap funding they lend against, and Slok said the speed and scale would be a problem for the entire financial system. JPMorgan Chase CEO Jamie Dimon flagged the same door on an October earnings call, warning against giving external agents unrestricted account access. 'You don't want rogue agents,' he said.
Edgars Nemše, chief executive of AI-focused blockchain project GenLayer Foundation, argues the banks may have a commercial motive they aren't saying out loud. Routine jobs — paying bills, moving money, hunting better rates — are exactly what agents are good at, and an agent can jump on a better deal the moment it hits the web. 'For plain accounts, that removes the switching costs and makes the space even more competitive,' Nemše tells Magazine. He also warns existing dispute processes were built for people making a few payments a week, not thousands of agent transactions per customer.
Others see agent-run banking arriving fast and mostly as an upgrade. Neeraj Makin, group head of strategy analytics at Emirates NBD, predicts agents will deliver around-the-clock banking and near-instant payments as the norm by 2036, with cross-border transfers settling in seconds rather than days and AI agents handling most financial decisions. Ferdinand Dabitz, chief executive of AI-native banking and payments firm Augustus, expects every consumer to run dozens of agents and every business hundreds, all talking to their bank — with a human interface still needed to manage them and hold the trust layer, 'especially in commercial banking.' Emirates NBD's developer team argued in July that institutions should build their own agentic frameworks rather than let individual teams loose; its Leap framework governs agent authentication, sensitive data access and actions, with audit trails and human oversight baked in.
The current reality is far more cautious. A recent Bank of England and FCA survey found 80% of responding firms had deployed AI agents, but only 2% used agents with broad autonomy, citing data security, third-party dependencies and hallucinations as top worries. In the US, the Office of the Comptroller of the Currency issued a supervisory framework on Nov. 4 covering banks' use of what it calls 'super intelligence,' including agentic systems. The framework carries no new binding requirements — it guides examiners on weighing an agent's autonomy and the level of human oversight against the consequences of its actions.
TopicsHSBC · Barclays · Lloyds · NatWest · Nationwide · JPMorgan Chase · Bank of England · Emirates NBD
Source: ct.com
Written by AI from verified reporting.

