There is a question that should be asked more often than it is.
When the FCA chose, in April 2024, not to write AI-specific rules, what kind of broker did that decision favour?
Most commentary read the Update as a technical clarification. It is more interesting than that. The FCA's choice of regulatory architecture, principles-based and outcomes-focused, is a directional bet on which firms can adapt. The bet is not neutral. It rewards certain kinds of broking and disadvantages others, regardless of size, segment, or product specialism.
This is what UK broking looks like under that bet, played out over the next five years.
The advantage flows to firms that can prove what they did
A principles-based regime measures firms on outcome and process. Outcome is the result. Process is the documented reasoning behind it. The FCA AI Update reinforces both. Section 3.43 commits firms to annual Consumer Duty assessments "evidenced with data." Section 3.40 confirms that any AI use sits within Senior Manager responsibilities, which require "reasonable steps to ensure that the business is effectively controlled." Section 3.32 requires explainability where automated decisions produce significant effects on consumers.
The common thread is documentation. Not technology. Documentation.
This favours firms that have built the muscle of writing things down. Recording the reasoning. Storing the evidence. Annual board reviews of outcomes. Systematic file notes. Audit logs of who decided what, when, on the basis of which inputs.
In 2030, the question a supervisor asks a broker will not be "do you use AI?" It will be "show me how you control it." Firms with mature documentation discipline answer that question in an afternoon. Firms without it answer it over six months of enforcement action.
Documentation is the dividing line. Not technological sophistication. Not capital. Documentation.
The advantage flows to firms that can explain their decisions
The Update treats explainability as central. Section 3.34 establishes that "AI systems should be appropriately transparent and explainable," and section 3.32 confirms that automated decisions with significant effect on consumers must include "meaningful information about the logic involved in the decision, as well as the significance and the envisaged consequences."
For UK brokers, this carries an implication most have not internalised. Whatever AI a broker uses, the broker remains accountable for explaining the decision to the customer, to the regulator, and potentially to the Financial Ombudsman Service.
Firms that build their workflows around explainability will scale into the next five years. Firms that adopt AI tools they cannot explain will hit a regulatory ceiling.
This favours brokers who treat AI as a co-pilot, not an oracle. The broker who can articulate why a particular product is suitable, citing the affordability data, the client's circumstances, the alternatives considered, and the reasoning that ruled them out, is operating within the regime. The broker who has accepted an AI recommendation without independent reasoning has handed their professional judgement to a tool whose logic they cannot reproduce.
The first kind of broker will own client relationships in 2030. The second will be defending complaints about advice they cannot remember giving.
The advantage flows to firms that govern AI as infrastructure
The Update positions AI use as a governance question, not a technology question. Section 3.13 brings AI inside operational resilience obligations. Section 3.15 brings it inside outsourcing rules. Section 3.40 places it inside SM and CR. Section 4.4 commits the FCA to deepening operational resilience scrutiny "increasingly relevant to firms' safe and responsible use of AI."
The implication is that AI tools in a broker's workflow are not features. They are critical third-party services. They require vendor due diligence, contractual protections, exit plans, incident response procedures, and impact tolerance assessments under SYSC 15A.
This sounds like overhead. It is competitive infrastructure.
Brokers who treat their CRM, their AI tools, their lead management systems and their compliance software as governed infrastructure will sell their firms in 2030 for materially higher multiples than brokers who treat them as a stack of free trials and one-person accounts. The reason is that the first kind of firm has built a transferable business. The second has built a personality cult around the broker, dependent on tools the broker controls personally and that disappear when the broker leaves.
Network principals will pay premiums for AR firms that demonstrate AI governance maturity. Lenders will accept introductions on better terms from brokers who can evidence Consumer Duty compliance through their AI use rather than despite it. PI insurers will price risk on the basis of AI governance, and the firms with documented frameworks will pay less.
These are not predictions. They are the existing economic logic of every regulated industry under principles-based supervision. The same dynamic that produced the consolidation of pension transfer advice, of structured product distribution, of equity release, will produce the consolidation of AI-enabled broking.
The brokers who get ahead of this are the ones who do the boring work in 2026 and 2027. The brokers who wait until the FCA writes a letter to the industry will pay materially more to catch up.
What this displaces
A principles-based AI regime makes some kinds of broking less viable.
It makes sole-trader broking dependent on consumer AI tools, with no governance documentation, hard to defend in supervisory engagement. It does not make it illegal. It makes it precarious in a way that scales with the volume of AI-influenced decisions in the workflow.
It makes networks that rely on AR self-reporting of AI use without verification, exposed at the principal level. Section 3.40 of the Update confirms this. Networks that do not implement firm-wide AI policies will discover their exposure when the first AR breach lands at the principal's door.
It makes opaque AI vendors increasingly unattractive. Brokers who cannot get a data processing addendum, an audit right, a SOC 2 report, or a meaningful service level agreement from an AI tool vendor will, over the next five years, find themselves uninsured, unrecommended by their networks, and unable to evidence compliance to their supervisors. The vendors who adapt will survive. The vendors who do not will lose the regulated market.
It makes credential-light advisory less defensible. The Update's emphasis on Senior Manager personal accountability under SM and CR (§3.40-3.41) means that the broker authorising AI-influenced advice carries that accountability personally. Brokers operating without the underlying technical qualifications to assess what their AI tools are doing will be exposed in a way the regulator has now confirmed it considers their problem.
What this favours
It favours brokers who treat their craft as continuous professional development. The Update's emphasis on explainability and oversight rewards advisers who can describe their reasoning rather than refer to a tool's output.
It favours firms with operational discipline. The same firms that already produce file notes, that already document their suitability reasoning, that already conduct outcome reviews, will absorb AI into their existing infrastructure with marginal additional friction.
It favours networks that build firm-wide AI policy. Networks that issue clear approved-tool lists, that vet AI vendors centrally, that audit AR compliance with the policy, will sleep better at night and price their risk lower.
It favours specialist brokers in defensible niches. The Update's outcomes focus means that the broker who genuinely understands their segment, the affordability of complex commercial deals, the realities of vulnerable client mortgage cases, the specifics of unusual property types, the underwriting nuances of complex protection cases, will not be displaced by AI. They will be amplified by it. Generic AI tools cannot replace specialist human judgement. They can make a specialist three times more productive while leaving the judgement intact.
It favours the brokers who read FCA documents.
That last point is not a joke. The FCA AI Update has been freely available for two years at the time of writing this article. Most UK brokers have not read it. The brokers who have, who understand the framework the FCA has chosen, and who have built their firms around it, are quietly compounding an advantage that will become visible in 2028 and decisive in 2030.
The bet the FCA has made
The FCA's choice not to write AI-specific rules is a statement of confidence in two things. That the existing principles-based regime is strong enough to absorb technological change. And that firms capable of operating within a principles-based regime are the firms the regulator wants to see thrive.
The bet does not protect brokers who want clarity. It rewards brokers who can produce clarity from a regulatory framework that demands judgement.
This is not the regulatory environment most UK brokers were trained in. It is the environment they now operate in. The next five years will reveal which firms made the adjustment.
The brokers who win will not be the brokers with the best AI. They will be the brokers who built the governance to use any AI safely, and who can prove it.