Artificial intelligence has become a routine source of financial information for younger British investors faster than many of them have understood where regulation stops. Research from the Financial Conduct Authority found that four in five less experienced investors aged between 18 and 40 have used AI to help with investing, while significant minorities incorrectly believe general-purpose AI-generated information carries regulatory or compensation protections.

The FCA surveyed 666 UK adults who either owned investments or were considering buying them during the following 12 months. Among that group, 56% said they trusted AI tools for investment information, putting them ahead of television and radio, the press, and social-media influencers.

Usage is expected to continue rising. Around two-thirds of respondents said they anticipated using AI more during the next year, while 38% considered it acceptable to make an investment decision solely from AI output. At the same time, 44% incorrectly believed AI-generated financial information was regulated, and 32% expected established compensation or ombudsman routes to protect them if AI advice caused a loss.

The figures expose an awkward boundary in Britain’s approach to AI and finance. General-purpose chatbots do not become FCA-regulated services simply because somebody asks them about shares, funds, or markets, whereas a system specifically designed and operated to provide regulated financial advice can sit much more clearly inside the regulator’s perimeter.

The interface can conceal the regulatory difference

The distinction depends partly on the activity being performed rather than whether the underlying technology is described as artificial intelligence. A broad chatbot can explain terminology, summarise information, or generate a view on an investment without automatically becoming an authorised adviser, while a regulated financial company configuring AI to provide covered recommendations remains responsible for the obligations attached to that service.

From the user’s perspective, the difference can be much less obvious. A polished conversational interface may respond in almost the same tone whether somebody is receiving general information from a public model or interacting with technology provided inside a regulated financial service, yet the routes for accountability and redress can differ considerably.

Respondents were not wholly unaware of AI’s weaknesses. Most recognised that systems can produce inaccurate information and that sources should be checked. The tension lies in simultaneously understanding that a model can be wrong while assuming a regulatory safety net exists behind the answer.

The immediate risk is not limited to conventional fraud. Generative systems can provide useful research, explain financial concepts, and organise public information while also producing hallucinated facts, stale data, or recommendations developed without a complete understanding of the person’s finances and objectives.

Regulated companies remain responsible for deployment

Within financial institutions, the FCA has taken a different approach from creating a separate rulebook for every use of artificial intelligence. It has said existing frameworks, including governance requirements and the Consumer Duty, continue to apply when regulated businesses introduce AI into activities already covered by financial regulation.

The regulator is testing that approach through initiatives including its AI Lab and live-testing work, where companies can explore deployment alongside regulatory and technical scrutiny. Such programmes examine the organisation using the system as well as the model itself, including governance, validation, accountability, and resulting customer outcomes.

Two conversations about investing can therefore carry markedly different obligations. A bank deploying AI inside a regulated process remains accountable for that service, whereas an individual asking a public chatbot which shares to buy may receive an equally confident response without the provider assuming the legal role of an authorised financial adviser.

The FCA’s survey suggests public understanding has not kept pace with that distinction. The finding that 44% believe AI-generated financial information is regulated indicates that some users are associating the subject of the answer with regulatory protection instead of examining who is providing the service and whether that organisation is authorised for the activity involved.

The boundary may become harder to recognise as models gain access to live data, persistent user profiles, and tools capable of taking actions rather than merely generating text. A system that remembers a person’s portfolio, income, previous decisions, and stated appetite for risk can feel increasingly like personalised advice even when its provider intends the product to remain a general-purpose assistant.

Financial businesses have their own incentive to use similar interfaces because conversational systems can reduce the cost of explaining products, handling support, and guiding customers through decisions. Inside a regulated environment, however, generating personalised language does not remove the company’s responsibility for governance, fair treatment, or the outcome of activities for which it remains accountable.

The FCA’s immediate response is therefore partly educational: investors should verify sources, retain their own judgement, and understand that the protections attached to regulated advice do not automatically follow information generated by a general-purpose model.

As conversational AI becomes a more ordinary interface for financial research, regulation will depend increasingly on whether people can tell which side of that boundary they have entered. The technology may look almost identical on the screen, while authorisation, accountability, and access to redress continue to depend on the organisation and service operating behind it.