FCA Floats Regulation of General-Purpose AI Models

The UK's Financial Conduct Authority (FCA) has suggested that Britain should consider regulating general-purpose AI models, such as those powering ChatGPT and Gemini. This recommendation was made by a senior official at the FCA on July 6, amid an ongoing debate about whether the UK's light-touch approach to AI governance is effective.

The FCA isn't proposing new AI-specific rules, but rather applying existing frameworks, like the Consumer Duty and Senior Managers and Certification Regime, to AI use cases in financial services. The regulator wants to regulate the underlying AI models themselves, not just how financial firms use them, which represents a subtle shift in their approach.

The recommendation comes from the broader Mills Review, launched by the FCA on January 27, 2026, to investigate how AI is reshaping retail financial services. The UK Treasury Committee has been critical of the FCA for lacking explicit guidance on AI usage in financial services, urging the regulator to clarify consumer protection and accountability rules by the end of 2026.

The UK has positioned itself as a pro-innovation hub for AI governance, unlike the EU's comprehensive AI Act. The FCA AI Lab and AI Live Testing pilot reflect this approach, allowing firms to experiment with AI technology in a supervised sandbox without triggering new legislative requirements.

Crypto firms using ChatGPT, Claude, or Gemini APIs for customer-facing financial advice or decision support could face new compliance obligations if the UK moves toward regulating general-purpose AI models. The accountability question also gets thorny, as it's unclear who would be responsible when an AI model hallucinates a piece of financial advice that causes consumer harm.