Skip to main content
Banks are banning AI agents over security... and competition concerns

Banks are banning AI agents over security... and competition concerns

Cointelegraph.com News
16 lis 2026, 13:37Cointelegraph
Oryginał
3 min read

Four major UK lenders, including HSBC and Barclays, have moved to block or restrict third-party artificial intelligence agents from accessing customer accounts, citing security and competition concerns.

The move comes as the banking sector grapples with the potential of AI agents to revolutionize how customers interact with their finances.

While there is currently no evidence of AI agents successfully breaching UK bank accounts, the lenders are acting preemptively to mitigate risks.

The primary concern is the potential for an automated system to mismanage funds, creating a complex liability issue regarding who would be responsible for any losses.

Competition and the 'Agentic Bank Run'

Beyond security, banks are wary of the competitive threat posed by AI agents that could autonomously find better financial deals for customers.

Edgars Nemše, chief executive of the AI-focused blockchain infrastructure project GenLayer Foundation, explains that agents could seamlessly identify and switch customers to more profitable accounts.

"For plain accounts, that removes the switching costs and makes the space even more competitive," Nemše tells Magazine.

Edgars Nemše, chief executive of GenLayer Foundation

This scenario could trigger an 'agentic bank run,' a concept warned about by Apollo chief economist Torsten Slok.

If swarms of agents move household cash from low-yield accounts to higher-return alternatives en masse, it would drain banks of the low-cost funding they rely on to make loans.

JPMorgan Chase chief executive Jamie Dimon has also voiced concern, stating that banks must avoid allowing external agents with unrestricted access.

"You don't want rogue agents," he said on an October earnings call.

Jamie Dimon, chief executive of JPMorgan Chase

Building the Future of Banking

Despite these short-term restrictions, industry leaders predict AI agents will fundamentally reshape banking within the next decade.

Neeraj Makin, group head of strategy analytics at Emirates NBD, forecasts a new era of around-the-clock banking and near-instant payments by 2036.

He envisions a future where 'all of that will be in seconds, atomic,' with AI agents taking over most financial decisions.

Ferdinand Dabitz, chief executive of the AI-native banking company Augustus, agrees that the customer interface will shift from direct human interaction to managing swarms of agents.

He notes that a human interface will remain critical for managing these agents and providing a layer of trust, particularly in commercial banking.

Regulatory and Operational Challenges

To manage this transition, some banks are developing their own frameworks for agentic AI, such as Emirates NBD's 'Leap' framework.

This framework is designed to govern how agents authenticate with core banking systems, access sensitive data, and carry out actions, incorporating audit trails and human oversight.

However, a recent survey by the Bank of England and the FCA found that while 80% of firms had deployed AI agents, only 2% used them with broad autonomy.

The survey highlighted data security, third-party dependencies, and risks like 'hallucinations' as major concerns.

The US Office of the Comptroller of the Currency issued a supervisory framework for banks' use of 'super intelligence' on Nov. 4.

The framework aims to help examiners assess how institutions use the technology, focusing on the potential consequences of an agent's actions, its level of autonomy, and the degree of human oversight.

Liability and the Path Forward

As AI agents become more capable, the question of liability when they make mistakes becomes increasingly urgent.

Edgars Nemše suggests that banks need to offer 'delegated access as a normal feature' to avoid customers sharing logins with agents.

This would involve customers setting permissions and spending limits, allowing agents to make low-value transactions independently while requiring approval for larger amounts.

Nemše warns that existing dispute processes may struggle if agents begin making thousands of transactions, as the current systems were built for human users making a few payments a week.

The integration of AI agents into banking is expected to happen in steps, starting with routine tasks like paying bills and subscriptions.

Banks that successfully navigate these challenges and develop robust governance frameworks are likely to come out on top in the evolving financial landscape.

Z.ai
Article written by AI
Z.ai GLM-4.7-FlashFree
Free AI articles150 of 150
No limit after signing in with Telegram