Can I use AI to research client investments and stay compliant?
Yes — if the AI cites its sources and the work is retained as a record. The compliance risk isn't AI itself; it's using a tool that can't show its sources or keep a books-and-records trail.
Short answer: Yes. Nothing in the Investment Advisers Act prohibits using AI for research. What the rules care about is whether you can (1) verify the basis for a recommendation and (2) retain a record of it. A general chatbot fails both; a sourced, logged research tool passes both.
Why the tool matters more than the technology
An RIA's obligations don't change because the input came from a model. Under the Advisers Act books-and-records rule, SEC Rule 204-2, advisers must retain records that form the basis of advice given to clients. If AI-assisted reasoning informs a recommendation, "it was in my chat history" is not a record you can reliably produce on exam.
So the question isn't "is AI allowed?" — it's "can this tool show its work and keep the receipt?"
The two-part test
Before AI output reaches a client, ask: (1) Can I trace every factual claim to a verifiable source? (2) Is this interaction retained in a system I could produce for a regulator? If either answer is no, don't stand behind the output.
What "compliant AI research" looks like in practice
- Citations on every claim. A figure from a 10-K should link to that filing, not appear as an unverifiable assertion.
- A retained trail. The query, the sources retrieved, and the answer are logged for SEC/FINRA retention — automatically, not by screenshotting a chat.
- Sources you'd actually cite. SEC EDGAR filings, economic data, market data, and your own firm documents — not the open web's best guess.
Where consumer chatbots fall down
They don't cite, they don't keep a compliant record, and they state wrong facts as fluently as right ones. That's fine for drafting an email; it's a problem the moment the output informs what you tell a client.
AdvisorIQ does research with citations on every claim and a full audit trail — built for the recordkeeping rules, not retrofitted to them.
Request access →General information, not legal or compliance advice. Confirm specifics with your own compliance counsel.
- Does the SEC ban AI for investment research?
- No. There is no rule prohibiting AI-assisted research. The applicable obligations — recordkeeping under Rule 204-2 and the duty of care — are technology-neutral and apply to the advice regardless of how it was produced.
- What makes AI research non-compliant?
- Using output you can't trace to a source, or that isn't retained as a record. An uncited claim that informs a recommendation, with no books-and-records trail, is the core risk.
- Is my ChatGPT history a compliant record?
- A consumer chat account is not a books-and-records system. If AI reasoning is part of a recommendation, you need it retained in a system you can produce on exam.
See a cited answer on your own questions.
Join the private beta. We're onboarding select firms now.
Keep reading
All q&a →What records must an RIA keep when AI informs a recommendation?
The same records you'd keep for any advice that forms the basis of a recommendation — plus the practical reality that AI output needs to live in a retainable system, not a consumer chat history.
Does the SEC Marketing Rule cover AI-generated content?
Yes — the Marketing Rule is about what you communicate to clients and prospects, not how you drafted it. AI-written marketing is held to the same substantiation and fair-and-balanced standards as anything else.
What's the difference between an AI notetaker and an AI research tool for advisors?
A notetaker captures what was said in the meeting. A research tool does the analysis before the meeting and the monitoring after it. They solve different problems — and only one cites sources you can stand behind.