The FCA's July 2026 review is not an overreaction. Four in ten UK consumers have used AI for financial advice, yet most of those systems have zero awareness of FCA Consumer Duty PS22/9, no grasp of individual circumstances, and no accountability when things go wrong. The regulator is now signalling that the current regulatory perimeter — which assumes a human adviser or at minimum a human oversight layer — no longer holds. This matters immediately to mid-market law firms, insurers, financial advisers and accountancy practices because the FCA will almost certainly tighten rules around autonomous AI agents making or recommending financial decisions. When that happens, firms still using generic off-the-shelf chatbots for client-facing work will face compliance gaps they did not know existed.

This story is part of a broader pattern. The EU AI Act came into force this year; the PRA has published guidance on AI governance (PRA SS1/23); the ICO continues to tighten UK GDPR enforcement against firms using third-party AI without clear data processing agreements. Regulators across the Commonwealth are watching. What the FCA is essentially saying is: you cannot outsource judgment to an LLM and claim you have met your regulatory duty. The old model — where AI was a research tool or internal efficiency play — is being overtaken by the new reality, where AI agents interact directly with consumers and make recommendations. The distinction between 'AI assistance' and 'AI autonomy' is no longer theoretical. It is regulatory.

Here is our honest take: generic chatbots like ChatGPT or Microsoft Copilot are not financial advisers. They cannot be. They have no view of a client's full circumstances, no understanding of regulatory perimeter, no accountability layer, and no audit trail that satisfies ICO UK GDPR or FCA expectations. Firms deploying systems like Harvey (for legal work) or Luminance (for document review) in a controlled environment with clear human oversight are on firmer ground — those tools augment specialist judgment, they do not replace it. But a retail financial services firm or an insurance broker pointing clients at a general-purpose AI chatbot for advice is now in direct conflict with what the FCA is about to mandate. The answer is not to ban AI. It is to build AI that understands regulatory context, maintains human oversight, logs decisions, and operates within a defined scope. That is the difference between reckless deployment and responsible integration.

What should you do now? First, audit any client-facing AI you are currently using. If it is a generic chatbot or an unmodified LLM, document the risks and escalate to compliance and leadership. Second, if you are planning to deploy AI for client interaction, insist on systems that integrate regulatory guardrails natively — systems that know what they are not permitted to do and log why they deferred to a human. Third, start monitoring FCA guidance actively. Trovix Watch is built for exactly this: tracking regulatory change in real time so you are not caught flat-footed when the new rules land. Fourth, if you are considering a client-facing AI assistant, ensure it is backed by an AI governance framework that satisfies ISO 42001 standards and can be audited. Trovix Audit gives you that visibility. The firms that survive the FCA's incoming tightening will be those that treated AI as a regulated tool from day one, not as a cost-saving experiment.

Source: City AM