The FCA says general-purpose AI tools are not regulated by it in the way that regulated financial advice from an authorised professional adviser is. Because it does not regulate advice from these tools, the FCA says users are not eligible for FSCS or FOS protections. In an FCA survey of 18- to 40-year-olds who own or are considering investments, almost half (44%) mistakenly believe AI-generated financial information is regulated.
The sources below come from the FCA, from ESMA, and from a joint article by the SEC Office of Investor Education and Advocacy, NASAA and FINRA. They discuss general-purpose AI tools and AI in investment services in general. None of them is presented here as an assessment of any specific product. Throughout, we separate what the sources state from our own interpretation.
What the FCA says about using AI tools for investing advice
This point appears in the FCA InvestSmart guide on using AI for investment research. The FCA says that general-purpose AI tools are not regulated by it in the way that regulated financial advice from an authorised professional adviser is.
The guide then sets out a consequence. The FCA does not regulate advice provided by general-purpose AI tools, and it says this means users are not eligible for protections from the FSCS or FOS.
The same guide also gives a name to one known problem: incorrect information provided by AI is known as "hallucinations". The sources cited in this article do not give a rate for how frequently that happens, so we do not state one.
Our interpretation: for a UK reader asking "can I use ChatGPT for investing advice?", the FCA's guidance points to two practical questions to keep separate. The first is whether the information is accurate. The second is what protection is available if something goes wrong. According to the FCA, the protections linked to regulated advice do not apply to advice from general-purpose AI tools.
How young UK investors view AI for investment research
The FCA press release on its investor survey reports research on 18- to 40-year-olds who own or are considering investments. The FCA compared how much these respondents trust different sources of information.
| Source of information | Share of respondents who trust it |
|---|---|
| AI tools | 56% |
| TV and radio | 47% |
| Press | 46% |
| Social media influencers | 29% |
The FCA also reported what the same group believed about the status of AI-generated information:
- Almost half (44%) mistakenly believe that AI-generated financial information is regulated.
- Around a third (32%) wrongly think they would get compensation from the FSCS or Financial Ombudsman Service if AI advice went wrong.
- Almost three quarters (73%) know that AI can provide inaccurate information.
The words "mistakenly" and "wrongly" come from the FCA's own description of these results. They are not our labels.
Who the FCA survey covered and when
How far these figures can be applied depends on who was asked. According to the FCA's description of its survey, the survey was conducted on July 24, 2026. The sample comprised 666 respondents based in the United Kingdom, and it was open to all adults across the 18 to 40 age range. The research focused on people in that age range who own or are considering investments.
Our interpretation: the results describe that group of respondents. They are not evidence about older investors, about investors in other countries, or about the general population. The survey measured what respondents said they trust and believe. It did not measure how accurate AI tools are or what happened to anyone's investments.
Comparing the survey results with the FCA's guidance
Read side by side, the two FCA documents address the same points from different angles. This is our own comparison, based only on what each source states:
- Regulation. In the survey release, 44% mistakenly believe AI-generated financial information is regulated. The InvestSmart guide says general-purpose AI tools are not regulated by the FCA in the way regulated advice from an authorised professional adviser is.
- Compensation. In the survey, 32% wrongly think they would get compensation from the FSCS or Financial Ombudsman Service if AI advice went wrong. The guide says users of advice from general-purpose AI tools are not eligible for FSCS or FOS protections.
- Accuracy. In the survey, 73% know that AI can provide inaccurate information. The guide notes that incorrect information provided by AI is known as hallucinations.
We do not draw a conclusion about why respondents hold these beliefs, because the sources do not give one.
AI-related investment fraud warnings from FINRA, NASAA and the SEC office
A separate concern is fraud that uses AI as a selling point. According to the joint investor article published by FINRA, the SEC Office of Investor Education and Advocacy, NASAA and FINRA jointly issued an investor article about investment frauds involving the purported use of AI.
Our interpretation: this source is about fraud schemes that claim to use AI, not about whether a general-purpose AI tool gives good or bad answers. These are separate risks, and this article does not treat one as evidence about the other.
ESMA guidance for firms using AI in investment services
The ESMA statement on firms using AI in investment services looks at the question from the firm's side. ESMA issued a Statement giving initial guidance to firms that use AI when they provide investment services to retail clients.
ESMA says it expects firms using AI to comply with relevant MiFID II requirements, including their obligation to act in the best interest of the client.
Our interpretation: this guidance is addressed to firms that provide investment services. As summarised here, it does not address individuals who put questions to a general-purpose AI tool on their own. We do not draw conclusions from it about any specific tool or provider.
Limits of this article
This article is educational information about what selected regulators and industry bodies have published on AI and investing. It is not investment advice and does not tell anyone what to buy, sell or hold. It does not evaluate any specific AI tool, measure how accurate AI answers are, or cover rules beyond the sources it cites. The survey findings describe young UK respondents and should not be applied to other groups.