The Financial Conduct Authority (FCA) has urged UK ministers to strengthen the City regulator’s powers to protect consumers from AI-driven risks [1].

This request comes as the watchdog identifies a critical gap in current oversight capabilities. As artificial intelligence integrates into the financial sector, the potential for sophisticated cyber-crime and fraud increases, threatening the security of consumer assets and data.

The FCA said there are amplified risks of cyber-crime and fraud [2]. The regulator is specifically concerned with how these technologies can be weaponized to bypass existing security protocols or deceive customers through highly convincing automated interfaces.

According to the Mills review, "companies are already starting to shift from human-led activitie" [3]. This transition toward automation reduces the human oversight that previously served as a buffer against certain types of systemic error and fraud.

The regulator is not only looking at immediate threats but is also analyzing how AI will reshape financial services from 2030 onward [1]. This long-term perspective suggests that the current regulatory framework may be insufficient for the structural changes expected over the next several years.

By requesting expanded powers, the FCA aims to create a more agile regulatory environment. This would allow the watchdog to respond more quickly to emerging threats without waiting for lengthy legislative updates, a necessity in the fast-paced evolution of generative AI and machine learning.

“"companies are already starting to shift from human-led activitie"”

The FCA's push for expanded authority signals a shift from a reactive to a proactive regulatory stance. By highlighting the transition away from human-led activities, the watchdog is acknowledging that traditional auditing and oversight methods are becoming obsolete. If ministers grant these powers, it could set a global precedent for how financial hubs manage the intersection of algorithmic trading, AI-driven customer service, and cybersecurity.