A Senior FCA Official Advocates Direct Regulation of AI Models in Britain
Sheldon Mills, an executive director at the Financial Conduct Authority (FCA), has proposed that Britain consider regulating large language models like ChatGPT, Claude, and Gemini directly. These models increasingly influence consumer financial decisions, but currently fall outside the regulatory perimeter.
Key Points:
- Concerns: Over 25% of UK consumers trust general-purpose AI tools for financial advice without realizing the lack of FCA protections. This raises questions about responsibility when such tools provide poor guidance.
- Proposed Remedy: Mills suggests that the FCA decide within 3-6 months whether to expand its regulatory perimeter to include these general-purpose models.
- Oversight Measures: He proposes new powers to require firms to explain their AI model decisions, audit algorithms for fairness, and impose fines for consumer harm caused by such systems.
Context:
- The UK has avoided a dedicated AI law, relying on existing regulators to oversee the sector, aiming for an innovation advantage over the EU.
- Mills’ proposal highlights a gap in this approach regarding general-purpose AI systems, suggesting the FCA may need to fill it.
- The debate about responsibility and regulation remains open, with direct model regulation versus focusing on regulated firms using them as options on the table.