Bank of England flags AI agent liability risks for financial stability

Traders Union
All News
Regulators
AI agent liability risks

Highlights

Bank of England Governor Andrew Bailey warns that AI agents in finance complicate legal liability, posing a significant financial stability risk discussed at the Group of Thirty event in Bangkok.
Wall Street banks have restricted AI agent access to client accounts due to concerns about shifting deposits, as regulators assess potential risks to liquidity and system stability.
G7 finance ministers and global officials, including the IMF and UK's FCA, highlight urgent cyber risk and transparency issues as AI models rapidly outpace current financial sector security controls.

Bangkok discussions sharpen focus on AI risks

As reported by Financial Times, Bank of England Governor Andrew Bailey says the growing use of AI agents in finance could mislead people over who carries legal liability when automated tools execute payments and other transactions on their behalf.

Speaking at a Group of Thirty event in Bangkok on Saturday, Bailey says regulators "have got to address" the issue. He describes the rapid development of AI as "a major issue" for financial stability and says the key question for authorities is how they should respond.

Bailey says there is already extensive discussion about regulating AI, but he argues that model testing is more important, partly to build public confidence that the technology is being developed responsibly. Describing what he calls the "agentic issue", he says people can create economic agents but not legal agents in most systems, making questions of responsibility, liability and risk management especially relevant.

Cyber, deposit and oversight concerns spread globally

Concern over AI in finance is already influencing bank policy and official planning. Several large Wall Street banks have restricted AI agents from accessing client accounts after raising concerns with regulators that the tools could shift idle cash in search of higher returns and drain deposits.

Officials in Bangkok also discuss broader systemic threats. G7 finance ministers this week consider simulated drills and tabletop exercises to guard against advanced AI models being used to launch a major cyber attack on the financial sector and global markets.

IMF Managing Director Kristalina Georgieva tells the same G30 event that the implications for financial stability could be very serious. She warns that without guardrails now, authorities may later face a form of non-human intelligence that is no longer controllable.

Recent warnings from Asia and the UK add to that pressure. South Korea's president says there are signs AI models have been used in attacks on banks, while the UK's Financial Conduct Authority warns that frontier models are finding cyber security flaws faster than financial services companies can fix them.

Shin Hyun-song, governor of the Bank of Korea, says AI remains highly opaque, calling it a very clever black box. Even so, he says regulators can still rely on traditional post-crisis tools such as capital requirements, leverage restrictions and minimum collateral levels because some basic rules remain relevant despite the uncertainty.

Our earlier article on the Bank of England’s push for bare metal recovery planning explained why banks are being urged to prepare for more severe tech failures as AI increases the scale and complexity of cyber threats. We noted Sarah Breeden’s warning in Bangkok that defences alone are no longer enough, and that lenders need independent recovery options to keep critical services running if core systems are compromised.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.