What happened
In June 2026, the UK Financial Conduct Authority published its first 'Emerging Technology Horizon Scan,' a forward-looking regulatory signal document covering three AI/technology convergence scenarios for financial services: (1) 'Personalised intelligence' — AI agents acting as delegated financial decision-makers, raising concerns about consumer agency, 'proxy economy' conflicts of interest, and AI-targeted dark patterns; (2) 'Synthetic (in)security' — AI-enabled financial crime, deepfake fraud, and synthetic identity attacks; (3) infrastructure/market-structure risks from technological convergence. This was accompanied by FCA CEO Nikhil Rathi's 24 June speech at techUK's Agents of Change event declaring the FCA wants to 'make it easy to safely build, test and scale AI in UK financial services' and that 80%+ of UK financial firms are already adopting AI. The FCA also re-opened its AI Input Zone (14 May 2026) to collect evidence of 'good and poor practice' across governance, resilience, deployment, oversight and assurance — explicitly seeking 'not theory, evidence.' On 15 May 2026, the Bank of England, FCA and HM Treasury issued a joint statement urging regulated firms to strengthen cyber resilience against frontier AI threats. The FCA is conducting live AI testing with Barclays, UBS, Lloyds, Experian and others through its supervised testing programme.
Why it matters
Although not a binding rule, the Horizon Scan and associated statements represent a material hardening of the FCA's supervisory posture on AI. The FCA's shift from principles to 'evidence of good practice' signals that the AI Input Zone submissions will feed into imminent good/poor-practice guidance with supervisory force under the Consumer Duty, SM&CR and operational resilience frameworks. The 'proxy economy' and AI-agent delegation scenarios preview the next wave of FCA AI governance concerns. UK-regulated financial firms should treat the AI Input Zone as a pre-enforcement feedback mechanism, not an academic consultation.
Action needed
Submit evidence to the FCA AI Input Zone while it is open — submissions will inform imminent good/poor-practice guidance with supervisory force. Map AI governance frameworks against Consumer Duty, SM&CR and operational resilience requirements now. Review the Horizon Scan's 'proxy economy' and agentic finance scenarios for forward planning. Ensure cyber resilience programmes address frontier AI threat scenarios per the BoE/FCA/HMT joint statement.