UK banks should be barred from some AI uses: researcher
Key Points
- New Durham University Business School research calls for a finance-specific AI rulebook based on the EU's AI Act.
- The framework sorts AI systems into four risk tiers, from unacceptable to minimal.
- Systems that manipulate or exploit customers fall into the unacceptable tier and should not be used at all.
- The research names the UK and US as light-touch jurisdictions compared with the EU and China.
- Consumer risks identified include misuse of personal data, bias, reliance on outside providers and cybersecurity gaps.
Finance firms should prohibit AI systems that manipulate or exploit customers outright, according to new research from Durham University Business School.
Habib Ahmed, Sharjah Chair in Islamic Law & Finance at Durham University Business School, authored the study, which sets out a global framework for how banks and other finance institutions should govern their use of AI.
Ahmed first mapped the risks the sector faces, finding that AI can affect core regulatory objectives including financial stability, consumer protection and financial integrity.
The consumer risks he identifies cover misuse of personal data, increased bias, over-reliance on third parties with no affiliation to the institution, and cybersecurity.
The framework borrows its structure from the European Union’s AI Act and sorts systems into four risk categories.
Unacceptable systems, including anything that manipulates people’s decisions or exploits them, such as scraping facial imagery, should be prohibited and not used at all.
High-risk systems, which can threaten health, rights or security or control access to vital services and data, need regulation and caution.
Limited-risk uses carry a transparency obligation, requiring providers to tell users when they are interacting with AI in content, text or video. Minimal-risk uses, including spam filters and video games, need no framework at all.
“AI laws are currently sparse and fragmented – but constantly evolving,” said Ahmed. “While some countries have implemented AI laws – like the EU and China – others have taken a lighter touch, like the UK and the US.”
Ahmed argues the stakes rise as adoption spreads across the sector.
“The use of AI in finance is going to increase in the future and its impact is set to be extremely high,” said Ahmed.
“Finance affects everyone – from the day-to-day consumer to the global markets – and if we are risking data leaks, biases in data or misleading information due to AI, we are drastically affecting all aspects of financial life, both at the micro and macro levels.”
Ahmed calls on policymakers and regulators to take frameworks such as the EU AI Act and apply them to finance directly.
