UK banks restrict third‑party AI agents from accessing customer accounts
Five major UK banks have moved to block or limit AI agents from making transactions, citing security, competition and systemic risk concerns.
Key takeaways
- HSBC, Barclays, Lloyds, NatWest and Nationwide have announced restrictions on third‑party AI agents accessing customer accounts.
- The banks cite security risks, potential “agentic bank runs” and competition concerns as reasons for the move.
- A recent Bank of England and FCA survey found 80% of firms use AI agents, but only 2% grant them broad autonomy.
Five of the United Kingdom’s largest banks – HSBC, Barclays, Lloyds, NatWest and Nationwide – have announced that they will block or severely limit third‑party artificial‑intelligence agents from accessing customer accounts, especially for autonomous transaction execution. The decision, reported by Cointelegraph, reflects growing worries about data security, the potential for “agentic bank runs” and the competitive impact of AI‑driven financial services.
Security and liability concerns drive the ban
Bank executives say the lack of clear liability when an AI agent mishandles funds is a key factor. Jamie Dimon of JPMorgan Chase warned on an October earnings call that “you don’t want rogue agents,” echoing the sentiment that banks could be held responsible for errors made by external AI systems. The move also aligns with a broader industry caution, as a Bank of England and FCA survey revealed that while 80% of firms have deployed AI agents, only 2% allow them broad autonomy due to fears of data breaches, hallucinations and third‑party dependencies.
Competition and the risk of agentic bank runs
Beyond security, the banks are wary of competitive erosion. Edgars Nemše, CEO of the GenLayer Foundation, told Magazine that AI agents could instantly compare rates and shift customers to better‑offering providers, eroding switching costs for plain‑account holders. Apollo chief economist Torsten Slok warned in September of a possible “agentic bank run,” where swarms of agents move cash from low‑yield checking accounts to higher‑return alternatives, draining banks of cheap funding needed for lending.
What regulators are doing
Regulatory bodies are beginning to address the technology. The U.S. Office of the Comptroller of the Currency released a supervisory framework for “super intelligence” on November 4, providing examiners with guidance on assessing AI‑driven systems, though it does not impose binding requirements. In the UK, Emirates NBD’s developer team has advocated for banks to create their own governance frameworks, such as the Leap model, which includes authentication controls, audit trails and human oversight for AI agents.
What to watch next
Industry observers will monitor whether other UK banks adopt similar restrictions and how fintech firms respond with alternative compliance solutions. The evolution of regulatory guidance, both in the UK and internationally, will shape the pace at which AI agents gain broader transactional authority. Finally, the emergence of “delegated access” features—allowing customers to set spending limits for agents—could become a compromise between automation benefits and risk mitigation.


Discussion 0