Is AI financial advice regulated? The FCA just published research that answers the question — and 44% of young UK investors have the wrong answer.
The survey, published 27 August 2026 and run on 24 July, covered 666 UK adults aged 18–40 who own investments or plan to buy them in the next year. Four in five have already used AI for help with investing. Almost half of them — 44% — mistakenly believe AI-generated financial information is regulated. It isn’t.
More than a third (38%) say it’s fine to make an investment decision based solely on AI output. And 32% wrongly expect the Financial Services Compensation Scheme (FSCS) or the Financial Ombudsman Service to pay out if AI advice goes wrong. They won’t. The FCA’s own words: “General purpose AI chatbots are not regulated.”
For agencies running AI-assisted ads for forex, crypto, and iGaming clients, this research is the demand side of your accountability problem. The regulator just quantified your client’s customers — and told you they can’t tell the difference between a regulated financial promotion and an unregulated chatbot.
The FCA just quantified your accountability problem
Read the survey again as an agency, not as a consumer. The 38% who would invest on AI output alone aren’t a hypothetical. They’re the people clicking your client’s ads. The 44% who believe AI-generated financial info is regulated aren’t confused about the technology — they’re confused about who’s responsible for it. The 32% who expect compensation don’t know there’s no safety net.
The FCA’s consumer-investments director, Lucy Castledine, put it plainly: “You need to understand how you’re protected and continue to use your own judgement.” The FCA’s own tip list includes “Know there’s no safety net” — “AI-generated tips from general-purpose chatbots mean you are not covered if things go wrong.”
Here’s the part that matters for your agency: financial promotions are regulated, and AI doesn’t change that. The FCA’s boundary line is one sentence: general-purpose AI chatbots are not regulated — but “tools which are specifically set up to provide financial advice would be likely to fall within the FCA’s remit.” An AI-assisted ad stack for a regulated vertical isn’t a general-purpose chatbot. It’s a tool built to move financial decisions. It sits on the regulated side of that line, which means the promotion it produces carries the full weight of the financial-promotion regime — including who approved it, and what record exists of it.
The trust inversion the FCA just proved
The survey found 56% of young investors trust AI tools more than TV and radio (47%), the press (46%), or social media influencers (29%). Read that again: the unregulated channel is the most trusted channel. The channels with actual regulatory oversight — broadcast, print, influencer rules — all score lower than a chatbot with no remit, no compliance function, and no record.
That’s the inversion that makes AI-assisted finance advertising dangerous: the more you automate, the more consumers trust the output, and the less protection exists behind it. Trust went up exactly where accountability went down. The FCA measured it; the number is 56% vs 29%.
The write era made this urgent
This is not a hypothetical-future post. The write era is here: Windsor has been writing campaigns from Claude since May, Optmyzr’s MCP writes, Supermetrics’ Claude integration creates campaigns, and seven new write-capable MCP servers launched in a single week (Aug 29–30). Meanwhile, the one competitor that claimed an audit trail on every write removed the claim nine days later (Aug 22 → Aug 31) — the moat vocabulary is borrowable; the moat is not.
Sixteen consecutive weeks of monitoring: zero finance PPC competitors with a compliance-capable write log. The market learned to write. Nobody shipped the write log. The FCA’s research is the demand-side proof that the gap has a body count: consumers are being told to “stay in the driving seat” by a regulator, while the AI-generated promotions they trust run through stacks that keep no record of who wrote what, who approved it, or when.
Who proves what the ad said?
The FCA’s research turns a product question into a regulatory one. When a prospect invests on AI output alone and it goes wrong, the question isn’t “was the chatbot regulated?” — it’s “who put that promotion in front of them, and what’s the record?” That’s the agency. The answer requires three things, each of which has a shipped answer:
- What did the AI generate, and what did a human approve? — Activity Logging records every change with actor attribution (human, tool, or AI agent by name), before/after values, and a timestamp. Not a dashboard snapshot — a per-action record, stored independently of Meta.
- When, where, and under which jurisdiction? — Agency Hierarchy tags every client → brand → account with jurisdiction labels (FCA, MiCA, ESMA, Alberta). When the FCA asks which promotion ran in the UK under which rules, the answer is a filter, not a forensic project.
- Can you export the record when asked? — A compliance-grade export in ~90 seconds, not a support ticket. Regulators don’t wait for your data team.
That’s the supply side of the FCA’s demand-side research. The regulator measured the belief gap; your record closes it.
The regulated funnel doesn’t stop at the ad
The FCA’s research describes the moment of trust — the click, the decision. But for your clients’ funnels, the decision is the start, not the end. The typical regulated-vertical path runs ad → Telegram community → KYC → deposit. Each hop is a place where AI output can appear: an AI-written ad, an AI-generated welcome message in the group, an automated DM to a prospect who “invested on AI output alone.”
The FCA’s survey found two-thirds of these investors expect to lean on AI even more over the next year. That’s not a trend line; that’s the direction of travel for your client’s entire funnel. Every AI touchpoint on that path is a representation that needs provenance — who wrote it, who approved it, when it went out, and under which jurisdiction’s rules.
That’s why Ott’s Activity Logging covers the whole funnel, not just the ad account: creative changes, campaign writes, budget moves, and Telegram-side actions all land in one timestamped, actor-attributed record. When the regulator asks what a prospect was told at each step, the answer is one export, not a reconstruction.
The questions your agency should be asked
Is AI financial advice regulated in the UK? General-purpose AI chatbots are not regulated by the FCA. Tools specifically set up to provide financial advice would be likely to fall within the FCA’s remit — and the FCA says so itself. For agencies, the relevant layer is the financial-promotion regime: AI-generated ads for regulated products are still promotions, and promotions are regulated.
What happens when AI advice goes wrong? The FCA’s answer: “You are not covered if things go wrong.” No FSCS, no Ombudsman, no safety net — which is exactly what 32% of young investors wrongly expect. The gap between expectation and reality is where the accountability question lands on whoever ran the promotion.
Does this apply to AI-generated ads, or just chatbots? The research measured consumer belief about AI-generated financial information — which is what ads are. The trust inversion (56% vs 29%) applies to the channel: the more AI-generated the promotion, the more it’s trusted, and the less regulatory protection consumers assume they need. That assumption is the exposure.
The 60-second agency check
- [ ] Can you produce, for any ad that ran last week, what the AI generated vs. what a human approved — with timestamps?
- [ ] Can you name the jurisdiction tag on every client account, and which promotion rules apply there?
- [ ] Can you export that record in under two minutes, without a data team?
If any answer is no, you’re running an unregulated channel’s trust with a regulated channel’s liability. The FCA’s research says 44% of your client’s prospects don’t know the difference. Your record is the only thing that does.