Consumer Duty and AI: what insurance brokers actually need to know in 2026

What Consumer Duty actually requires brokers to evidence

Consumer Duty is built around four specific outcomes that the FCA now actively monitors:

Products and services: Are the policies you place designed to meet identified client needs? Are you distributing them to appropriate target markets?

Price and value: Is there a reasonable relationship between what the client pays and the benefit they receive? Can you justify your commission and fee structure?

Consumer understanding: Do your communications, from pre-placement summaries to renewal reports, support clients in making informed decisions?

Consumer support: Can your clients access the help they need throughout the policy lifecycle, including at claim?

The shift in 2026 is that the FCA expects evidence against all four, not just policies and procedures. Browne Jacobson's 2026 horizon scan puts it plainly: brokers must clearly evidence how they are delivering good customer outcomes in practice. Board minutes matter. Decision records matter. The FCA is distinguishing between firms that actively engage with outcome data and those that merely report it.

For brokers, this means every placement decision, every comparison, and every piece of client advice needs a traceable trail.

What the FCA is signalling about AI in 2026

The FCA does not plan to introduce AI-specific regulation. Their published approach remains principles-based and outcomes-focused. Existing frameworks — Consumer Duty, SM&CR, and operational resilience rules — already apply to any AI system a broker deploys.

That said, the regulator is watching closely. In January 2026, the FCA launched the Mills Review, a long-term examination of how AI will reshape retail financial services by 2030. Led by Executive Director Sheldon Mills, the review covers agentic AI, market structure implications, and the possibility that consumers will increasingly interact with financial services through AI-mediated interfaces rather than directly with firms.

The Treasury Committee has separately recommended that the FCA publish comprehensive AI guidance for firms by the end of 2026.

What this means for brokers right now is straightforward: the FCA will not tell you which tools to use. But they will hold you accountable for the outputs those tools produce. The responsibility sits with the broker, not the technology vendor. If you use an AI system to compare policy wordings or draft a client summary, you need to be able to explain how it reached its conclusion and trace it back to source material.

The gap most independent brokerages have right now

Most brokers do not miss things because they lack expertise. They miss things because they are reviewing lengthy policy wordings late in the day, moving between renewals, endorsements and insurer responses at speed.

Anyone who has worked through a January renewal season knows how quickly good process starts to strain under volume.

This is where manual workflows fail against Consumer Duty requirements. Consider the four outcomes in practice:

Products and services: You compared three PI wordings. Can you show which exclusions differed? Can you evidence why you recommended Insurer A over Insurer B?

Price and value: You placed a client with the cheapest quote. Can you demonstrate the cover was adequate, not just affordable?

Consumer understanding: You sent a renewal summary. Was it clear enough for the client to understand the material gaps between this year's cover and last year's?

Consumer support: A client called about a circumstance notification clause. Can you retrieve what you told them, and when?

In a five-person team handling 200 renewals in January, the answer to most of these is: it depends on who did it and how much time they had.

That inconsistency is the real compliance risk. Not bad advice, inconsistent documentation of good advice.

What good looks like: AI with an audit trail

AI becomes relevant here not as a shortcut, and not as a replacement for broker judgement, but as infrastructure that makes good practice sustainable at volume.

There is a version of this story that concerns regulators: brokers using tools they do not fully understand, relying on outputs they cannot verify, introducing more opacity into the advice process. The FCA has been clear on this. When they talk about supporting AI adoption, they mean systems firms can explain.

For a broker evaluating AI tools, the test is practical. Any system you deploy should meet these five criteria:

1. Grounded in your documents, not general knowledge The AI should work from the actual policy wordings, schedules and endorsements you upload, not internet-sourced summaries. If the answer is not supported by the document, the system should say so.

2. Traceable outputs Every comparison, summary or recommendation the AI produces should reference the specific clause, page or section it drew from. A regulator should be able to reconstruct the reasoning months after the event.

3. Timestamped records When the comparison was run, what documents were included, what the output was, and whether a broker reviewed it. This is your audit trail against Consumer Duty.

4. Broker review at every decision point The AI drafts. The AI flags. The broker decides. If there is no human review step before outputs reach a client, you have an accountability gap under SM&CR.

5. Data segregation and security Client documents should not be used to train models. Data handling should be GDPR-compliant. If your team is pasting policy schedules into general-purpose AI tools, that is both a data protection risk and a Consumer Duty risk.

This is not new, it is just harder to do manually

A broker comparing two policy wordings, documenting the differences, explaining material gaps to a client and retaining a record of that conversation is already operating in line with Consumer Duty principles. That process has always been what good broking looks like.

The issue is whether it happens every time. Whether it happens at 4pm on a Friday in January the same way it happens on a quiet Tuesday in March.

The FCA's 2026 supervisory approach makes the stakes of that inconsistency explicit. Firms that consistently deliver good outcomes face less intensive supervision. Firms where the FCA identifies harm will face quicker and more decisive intervention.

The brokerages treating compliance as an operational discipline, embedded in the renewal and advice process itself, not layered on top as documentation after the fact, are in a far stronger position for whatever the FCA asks next.

What brokers should do now

Audit your current documentation trail. Pick five recent renewals at random. For each one, can you evidence the comparison process, the placement rationale, and what the client was told about material differences? If the answer varies by handler, you have a consistency problem.

Review any AI tools your team is already using. If anyone is using ChatGPT, Copilot or similar general-purpose tools on client work, assess whether those outputs are traceable, document-grounded and retained. If not, that is an exposure.

Evaluate purpose-built alternatives. AI tools designed specifically for broker workflows, grounded in uploaded policy documents, with citation and audit trail built in, exist now. The question is not whether to adopt AI. It is whether the AI your team is already using meets the standard the FCA expects.

Assign Consumer Duty ownership. The FCA expects a Consumer Duty Champion at board level. If you have not done this, or if the role is nominal, fix it. That person should be reviewing outcome data, not just signing off documents.