The FCA's call for expanded powers to regulate AI is not ideological posturing. It reflects a concrete problem: UK consumers are increasingly asking ChatGPT, Perplexity and other general-purpose models for financial guidance, and those systems have no concept of FCA Consumer Duty PS22/9, no understanding of suitability, and no accountability when they get it wrong. The regulator wants to expand its perimeter to cover autonomous AI models and build a public-interest AI guidance service. For mid-market law firms, insurers, financial advisers and accountancy practices, this means one thing: if you have deployed AI into client-facing workflows or internal advice generation, you are now operating in a space the FCA is actively scrutinising. The days of 'we bought an LLM and bolted it onto our website' are ending.
This story is part of a broader pattern we are watching across UK regulated sectors. The EU AI Act is live. The ICO is tightening GDPR enforcement. The PRA and FCA are aligning on AI governance expectations. Lloyd's Blueprint Two now requires underwriters to audit algorithmic decision-making. The pattern is clear: regulators have moved from 'we don't understand AI' to 'we do, and you need to prove yours is safe.' The difference between a firm that gets ahead of this and one that gets caught flat-footed is not technical sophistication—it is governance rigour. Generic AI tools like Microsoft Copilot, ChatGPT, and Claude are powerful, but they are not built for regulated advice. When they hallucinate, your firm takes the hit, not OpenAI.
Our view: deploying AI without governance infrastructure is regulatory risk wearing a productivity mask. The right approach is not to avoid AI—it is to build AI systems with four things baked in from day one: transparent decision logic (what did the model actually use to reach that conclusion?), audit trails (who asked what, and what did the system return?), human checkpoints (is there a qualified person validating the output before it reaches a client?), and guardrails specific to your regulatory obligations (not generic chatbot behaviour, but FCA-compliant, SRA-compliant, or FRC-compliant reasoning). Tools like Trovix Audit exist precisely because firms need to know what their AI systems are doing in real time, not discover it in a complaint letter. That gap between 'we deployed it' and 'we understand it' is where FCA enforcement actions happen.
What should a regulated firm do now? First, audit what you have deployed. If you have bots, copilots, or autonomous systems touching client data or generating advice, document them. Second, establish a baseline: what regulatory obligations does each system have, and how is it meeting them? Third, build or buy governance infrastructure—not as an afterthought, but as the foundation. If you are a financial services firm, use Trovix Watch to track FCA rule changes in real time so you are not three months behind when new guidance lands. Fourth, invest in explainability. The FCA will not approve systems that work like black boxes. Fifth, involve your compliance and legal teams before you roll out the next AI pilot, not after a complaint arrives.
Source: City AM