UK regulator warns of "arms race" to keep up with AI use in financial services

A senior UK financial regulator has described the effort to oversee artificial intelligence in financial services as an “arms race,” warning that current powers are insufficient as millions of consumers turn to AI tools for personal finance decisions. Sheldon Mills, an executive director at the Financial Conduct Authority (FCA), told the Financial Times that regulators would need greater powers to keep pace with the rapid adoption of AI [1]. The FCA regulates roughly 58,000 businesses in the UK financial sector, which contribute around £65.6 billion in annual tax revenue [10]. Mills urged authorities to review whether large language models such as ChatGPT, Claude, and Gemini should fall under existing rules [1]. “It is an arms race,” he said [1]. The warning coincides with the publication of an FCA-commissioned report authored by Mills on AI’s impact in financial services [1]. The report recommends that the FCA carry out a review within three to six months to examine the risks posed by companies offering financial services outside the regulator’s remit, as well as potential consumer harm from the growing use of AI models to manage personal finances [1]. Research cited in the report found that one-fifth of UK adults were already open to using AI to make financial decisions on savings or borrowing, despite those services lacking regulatory coverage and any path to compensation if errors occur [1]. The push for updated oversight reflects a broader global challenge. The regulatory landscape for AI remains an emerging issue across jurisdictions, with the European Union adopting its AI Act in 2024 to establish a common legal framework [9]. The FCA, which operates independently of the UK government and is funded by industry fees, has been led since 2020 by chief executive Nikhil Rathi, the former CEO of the London Stock Exchange [10][11]. Mills noted that some firms have described AI-driven services as economically equivalent to regulated advice while sitting outside the regulatory perimeter, where “reasonably strict” rules apply to comparable human recommendations [1].