AI Regulation Tracker / Regulator guidance
UK FCA Keeps Named Senior Managers Personally Answerable for AI Outcomes
The FCA's Mills Review, published July 6, 2026, confirms it will not write AI-specific rules. Instead a named senior manager stays personally accountable for AI-driven outcomes under the existing Senior Managers regime.
The UK's Financial Conduct Authority has decided that the arrival of more autonomous AI does not change who is on the hook when a model harms a customer. In the Mills Review, published on July 6, 2026 and led by FCA executive Sheldon Mills, the regulator confirmed it will keep its existing accountability framework and apply it to AI rather than build a separate AI rulebook.
What the review actually says
The review, titled "AI and the future of retail financial services," carries 7 priority recommendations for the FCA Board and Executive. Its central message for regulated firms is a choice not to act in one specific way. The FCA will not write AI-specific rules. It will instead hold to what its Chair, Ashley Alder, described as a "principles-based, outcomes focussed approach" that relies on the Consumer Duty and the Senior Managers Regime. The regulator has committed to issue clearer guidance on the reasonable steps a senior manager is expected to take when AI sits inside a regulated process.
Personal accountability does not move
The load-bearing point for compliance teams is continuity. The Senior Managers and Certification Regime continues to apply as AI systems become more autonomous. A named senior manager remains personally answerable for AI-driven outcomes, even where the behaviour of a model sits partly outside the firm's direct control. That accountability does not pass to the software vendor, the model provider or the model itself. If an AI system produces an unfair outcome for a retail customer, the FCA can look to a specific human being who was responsible for overseeing it.
This is a deliberate design. Rather than defining new categories of AI liability, the FCA is extending a regime that already names accountable individuals and asks whether they took reasonable steps. The practical burden is documentation. A senior manager who owns an AI system needs a record showing the controls, testing, monitoring and escalation routes that were in place, so that "reasonable steps" can be evidenced rather than asserted.
The phrase "reasonable steps" is doing a lot of work. In an ordinary process, a senior manager can point to sign-offs, second lines of defence and clear reporting lines. With a model that adapts over time, the same manager has to show ongoing oversight, not a one-off approval at launch. That is why the review's promise of clearer guidance matters. Until it arrives, firms are left to interpret an existing standard against a moving technology, and the safe reading is to over-document rather than under-document.
What it does not do
The review does not create AI-specific offences, penalties or a licensing regime for AI systems. It does not set numerical accuracy thresholds or ban particular models. It does not shift the framework toward prescriptive technical rules. Firms hoping for a detailed checklist will not find one here. The trade-off is familiar to anyone who works under principles-based regulation: more flexibility, but also more responsibility to interpret and justify decisions when the regulator asks. The review is also a set of recommendations to the FCA's own Board and Executive, so some of its content is about how the regulator will supervise, not a fresh list of firm-level obligations that took effect on publication day.
The cross-border angle for US readers
US financial groups with UK retail operations fall squarely within the Senior Managers and Certification Regime, so this is not a distant foreign development for compliance leaders in New York or Charlotte. It also reads as a signal. A major regulator has chosen to stretch an existing accountability regime over AI instead of inventing new AI-specific liability. US firms should expect the same instinct elsewhere, where supervisors reach for the governance and officer-accountability tools they already have rather than wait for bespoke AI statutes. The named-owner model the FCA is reinforcing is a reasonable template to adopt voluntarily now.
What to do this quarter
The concrete task is ownership. Firms should confirm which senior manager is accountable for each material AI system, ensure that responsibility appears in the relevant statement of responsibilities, and build the evidence file that shows the oversight steps taken. Doing that ahead of the FCA's promised guidance puts a firm in a defensible position whatever the detail of that guidance turns out to be.
Frequently Asked Questions
What changed with the FCA Mills Review?
On July 6, 2026 the FCA published the Mills Review on AI in retail financial services, with 7 priority recommendations. It confirmed the FCA will not write AI-specific rules and will instead apply its existing Senior Managers regime to AI, with clearer guidance on the reasonable steps senior managers must take.
Who is affected?
Senior managers and FCA-regulated retail financial services firms in the UK, including banks, insurers, financial advisers, consumer-credit firms and payments firms. US financial groups with UK retail operations are covered by the same Senior Managers and Certification Regime.
Does accountability shift to the AI vendor or the model?
No. A named senior manager remains personally answerable for AI-driven outcomes, even where a model's behaviour sits partly outside the firm's direct control. Responsibility stays with the accountable individual, not the software provider or the model.
Are there new AI-specific penalties or rules to comply with?
No new AI-specific rulebook was created. The FCA is keeping its principles-based, outcomes-focused framework and applying existing obligations, including the Consumer Duty and the Senior Managers Regime, to AI systems.
Sponsored Training
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.