FCA — Mills Review findings and recommendations (published 6 July 2026)

Tag: S-2026-07-06-fca-mills-review-findings Type: report (independent review commissioned by the FCA) Author(s): Financial Conduct Authority — review led by Sheldon Mills (Executive Director, Consumers and Competition) Date of source: 2026-07-06 (findings and recommendations published; FCA hosting a same-day event) Date ingested: 2026-07-06 Authority weight: high — the UK conduct regulator’s own long-term review, now in published form. Provenance caveat: the FCA publications/news pages were not directly retrievable this run (JavaScript-rendered / provenance-restricted); findings confirmed via WebSearch of fca.org.uk and multiple legal-firm summaries (Freshfields, Baker McKenzie, regulationtomorrow.com). Exact recommendation wording to confirm against the FCA’s own published document. Raw file: S-2026-07-06-fca-mills-review-findings

What it claims

The Mills Review — Review into the long-term impact of AI on retail financial services — was published on 6 July 2026, launched 27 January 2026 and led by Sheldon Mills. It looks beyond current use cases to how increasingly advanced, autonomous and interconnected AI systems could reshape market structure, firms’ operations, consumer behaviour and regulatory approach by 2030 and beyond — a landscape it expects to be defined by systems that are more autonomous, adaptive and interconnected, moving toward agentic AI capable of independent decision-making and continuous learning.

The review’s central recommendation is that the FCA should publish, by the end of 2026, comprehensive and practical guidance for firms covering (i) the application of consumer protection rules to firms’ use of AI, and (ii) the accountability and level of assurance expected from senior managers under the Senior Managers and Certification Regime (SM&CR) for harm caused through the use of AI.

The review reaffirms the FCA’s settled stance that it does not intend to create an entirely new AI rulebook. Instead the FCA aims to assess whether existing regulatory frameworks — the Consumer Duty, SM&CR and operational resilience requirements — remain appropriate in an AI-enabled environment. The review invited input across four themes: the future evolution of AI technology (including the full AI value chain and more autonomous/agentic systems); the impact on markets and firms (competition, market structure, UK competitiveness); the impact on consumers (outcomes, new risks, changing behaviours, demand); and the future shape of regulation.

Notable quotes

• Review framing (Sheldon Mills): “designing for the unknown” — carried from the launch speech (28 Jan 2026); a verbatim quote from the published findings was not captured this run.

What’s speculative vs. asserted

• Asserted (review position): publication on 6 July 2026; the by-end-2026 guidance recommendation on (i) consumer-protection-rule application to AI and (ii) SM&CR senior-manager accountability/assurance for AI harm; the “no new AI rulebook — test existing frameworks” stance; the 2030+ agentic-AI horizon framing; the four consultation themes.
• Forward-looking: whether and how the FCA adopts the recommendation; the content and exact timing of the end-2026 guidance; whether any change to Consumer Duty / SM&CR / operational resilience levers follows.
• Provenance caveat: findings confirmed via WebSearch and secondary legal commentary, not a directly-retrieved FCA page this run. Treat exact wording as medium-high confidence pending confirmation against the FCA’s published document.

Topics this feeds

• FCA approach to AI — converts the anticipated 6 July publication into confirmed findings; the by-end-2026 guidance recommendation is the key new datum.
• Model Risk Management and Agentic AI — the review’s agentic/autonomous-AI-to-2030 framing and SM&CR-accountability recommendation reinforce the “where the framework strains” and accountability threads.
• FCA — entity timeline update.

Open questions raised

• Will the FCA formally adopt the by-end-2026 guidance recommendation, and will that guidance introduce concrete, testable benchmarks for “good vs poor practice” (linking to the AI Input Zone submissions closed 19 June 2026)?
• Does the review recommend any actual change to the Consumer Duty / SM&CR / operational resilience levers for AI, or affirm their sufficiency for agentic/autonomous systems?
• How will “the level of assurance expected from senior managers” under SM&CR for AI-caused harm be evidenced in practice — the direct hook for an independent-assurance offer?

Ingestion note

Supersedes the timing-only capture S-2026-06-26-fca-mills-review-date (the 6 July date is now confirmed by publication; that source is preserved, not deleted). Captured via WebSearch fallback — the FCA news/publications pages were not directly fetchable this run. The publication date and the by-end-2026 guidance recommendation are corroborated across multiple independent legal-firm summaries; exact recommendation wording should be reconciled against the FCA’s own published document on a future run with direct retrieval.