AI in action · Regulation
What US Regulators Expect From Brokers Using AI Agents
There is no special rulebook for AI in US brokerage. FINRA's 2026 oversight report applies the existing one, and for the first time spells out what can go wrong with generative AI and autonomous agents.
As brokers hand more work to AI, the question of who is responsible when it goes wrong is becoming practical. In the US, the answer for now is: the same firms, under the same rules.
FINRA, the self-regulatory body that supervises US broker-dealers, included a dedicated section on generative AI for the first time in its 2026 Annual Regulatory Oversight Report, published in December 2025. Its starting point is that its rules are "technology neutral": supervision, communications with customers, record-keeping and fair dealing apply to an AI tool exactly as they would to any other system. The report itself says it does not create new requirements.
It does describe the risks in detail. The most common use among member firms is summarising documents and extracting information. The report defines hallucinations as outputs that are inaccurate or misleading — a misread rule, or wrong client or market data that could affect a decision — and asks firms to have procedures to catch them. Suggested controls include keeping logs of prompts and outputs, recording which model version was used and when, and having humans review results, according to WealthManagement.com.
AI agents get their own warning. FINRA describes them as systems that can plan, decide and act autonomously, and flags the risk that they act beyond the scope and authority the user actually intended. Supervision, it says, may need to be tailored to the type and scope of each agent.
The Securities and Exchange Commission has taken a similar route. It withdrew its 2023 proposal on predictive data analytics in June 2025 and has no AI-specific rule for advisers. Instead, its examination priorities for fiscal 2026 include "AI washing" — misleading claims about how much a firm really uses AI — and how firms supervise AI used in trading and portfolio management. The SEC settled charges over exactly that kind of claim with two advisers, Delphia and Global Predictions, in March 2024.
For individual investors, the practical point is simple: a regulated broker remains responsible for what its AI tools do, but a general-purpose chatbot you use on your own is not supervised by anyone.
Sources: FINRA — 2026 Annual Regulatory Oversight Report (Dec 2025); WealthManagement.com — FINRA cautions broker/dealers on gen AI hallucinations (Dec 9, 2025); Goodwin — 2026 SEC exam priorities for investment advisers