
UK Banks Block AI Agents From Accessing Customer Accounts
This article was produced with AI assistance and reviewed for accuracy.
Content is for informational purposes only and does not constitute financial advice.
Event: Five UK Lenders Restrict Third-Party AI Agents
HSBC, Barclays, Lloyds, NatWest and Nationwide have all moved to block or restrict third-party AI agents from accessing customer accounts, especially where agents could make transactions without customer involvement, Cointelegraph reports. There is no evidence yet of AI agents actually accessing UK bank accounts, but the lenders are moving pre-emptively. The core concern is accountability: if an automated system mismanages a customer's money, it is unclear who would be responsible. Edgars Nemše, chief executive of GenLayer Foundation, an AI-focused blockchain infrastructure project, told Cointelegraph Magazine that routine banking tasks such as paying bills, moving money and finding better rates could all realistically be handed to agents. He noted that this capability cuts against the banks' commercial interests because agents "would be able to seamlessly find customers a better deal elsewhere as soon as it's published on the web," adding that "for plain accounts, that removes the switching costs and makes the space even more competitive."
Cointelegraph suggests the banks may have other, unspoken concerns too, and competition may be among them. If agents can seamlessly find customers a better deal elsewhere, the switching costs that keep many customers with their existing bank fall away, a scenario the report says could threaten more than customer loyalty.
The risk is not purely commercial. In September, according to Cointelegraph, Apollo chief economist Torsten Slok warned of a possible "agentic bank run." If large numbers of AI agents began moving household cash out of low-yield checking accounts into higher-return alternatives simultaneously, that could drain banks of the low-cost deposit funding they rely on to issue loans — and Slok said the speed and scale of such a shift "would be a problem for the entire financial system." JPMorgan Chase chief executive Jamie Dimon has also pushed back on giving external agents unrestricted account access, telling an October earnings call, "You don't want rogue agents."
Implications: Regulators and Frameworks Are Already Responding
The US Office of the Comptroller of the Currency issued a supervisory framework on Nov. 4 covering what it terms "super intelligence," including agentic systems, per Cointelegraph. The framework does not introduce new binding requirements but is intended to help bank examiners judge how institutions deploy the technology — with scrutiny calibrated to an agent's level of autonomy, the potential consequences of its actions, and the degree of human oversight involved. Cointelegraph's reporting draws the distinction clearly: an AI system helping a bank employee draft a report carries a different risk profile from one authorized to move customer funds without approval.
Emirates NBD's developer experience team argued in July that financial institutions should build their own governance frameworks for agentic AI rather than letting individual teams build agents without shared controls. Its "Leap" framework, as described to Cointelegraph, is designed to govern how agents authenticate with core banking systems, access sensitive data and carry out actions, with audit trails, monitoring and human oversight built in from the start.
Survey data cited in the Cointelegraph report underscores why caution dominates for now. A Bank of England and FCA survey found that while 80% of responding firms had deployed AI agents in some form, just 2% used agents with broad autonomy. Firms identified data security, third-party dependencies and risks such as hallucinations among their biggest concerns. That gap between broad experimentation and limited autonomy is the backdrop to the UK restrictions.
CoinHours Analysis: Why This Matters for Crypto-Adjacent Infrastructure
CoinHours analysis: this is not a crypto story in the narrow sense, but it sits inside the broader debate over autonomous agents handling money, which has clear parallels with crypto wallets and smart contracts that can move value without a bank in the loop. Nemše told Cointelegraph that existing dispute processes may struggle if agents make thousands of transactions per customer: "Even a small share going wrong is a lot of disputes," he said, adding that "the processes for handling them were built for people making a few payments a week. That part still has to be built."
The gradual-adoption path Nemše describes to Cointelegraph — starting with agents that queue payments for human approval, before eventually being trusted to spend small amounts autonomously on bills, subscriptions and their own running costs — resembles the permission and spending-limit controls he says banks should offer as "delegated access as a normal feature", so customers do not share logins with agents. Ferdinand Dabitz, chief executive of Augustus, told Cointelegraph that "every consumer will have 10s and every business will have 100s of agents talking to their bank," while still expecting human interfaces to remain important "especially in commercial banking." Neeraj Makin, group head of strategy analytics at Emirates NBD, predicted that near-instant payments will be normal by 2036, with cross-border transfers that now take days settling "in seconds, atomic" as agents take over most financial decisions.
What to Watch
The near-term signal to track is whether other major lenders beyond HSBC, Barclays, Lloyds, NatWest and Nationwide follow with their own restrictions, and whether the OCC's Nov. 4 framework evolves from a supervisory assessment tool into something closer to binding rules. Dabitz told Cointelegraph that corporate treasury management could be an area delegated to agents, but said banks "will need to be involved in developing these early applications." The dispute-handling infrastructure gap Nemše flagged — built for a world of a few manual payments per week, not thousands of agent-driven transactions — still has to be built, in his words, before agents can act at scale.
Further reading / context
Frequently Asked Questions
Which UK lenders are restricting AI agents?+
HSBC, Barclays, Lloyds, NatWest and Nationwide have moved to block or restrict third-party AI agents from accessing customer accounts, especially where agents could make transactions without customer involvement.
Why are UK banks restricting AI agent access to customer accounts?+
The main stated concern is accountability: it is unclear who is responsible if an automated system mismanages a customer's money. Cointelegraph also points to competition, since agents could find customers better deals instantly and remove switching costs.
What is an 'agentic bank run' and why is it a concern?+
According to Apollo's chief economist, it's a scenario where large numbers of AI agents simultaneously move household cash from low-yield checking accounts into higher-return alternatives, potentially draining banks of low-cost deposit funding needed for lending.
What regulatory framework has been established for AI agents in banking?+
The US OCC issued a supervisory framework on Nov. 4 for 'super intelligence' systems. It doesn't create binding requirements but helps bank examiners assess how institutions deploy agentic technology based on autonomy level, action consequences, and human oversight.
What percentage of UK financial firms currently use AI agents with broad autonomy?+
A Bank of England and FCA survey found that while 80% of responding firms deployed AI agents in some form, only 2% used agents with broad autonomy, citing data security, third-party dependencies and hallucination risks.
Sources & Further Reading
Get CoinHours in your inbox
Crypto markets, cycles and on-chain signals in your inbox.

