Best AI for AdvisorsWhat is a hallucination?
Also called: AI hallucination.
Short answer
When an AI tool states something false or made up as if it were fact, such as a wrong figure, rule or citation.
Definition
A hallucination is confident, wrong output. The AI might invent a tax rule, misstate a client's balance from a statement, attribute a quote to the wrong person or cite a source that doesn't exist. It happens because generative models predict plausible text rather than look up facts.
FINRA's 2026 oversight report defines hallucinations as instances where the model generates information that is inaccurate or misleading yet presented as factual.
Why it matters for advisors
In an advisory firm a hallucination can become a wrong number in a client email, an inaccurate meeting summary in the file, or a misstated rule in a compliance memo. The SEC Marketing Rule and FINRA Rule 2210 don't care whether a person or a tool wrote the untrue statement.
That's why the standard control is human review: an advisor checks AI output against the source before it's used, especially numbers, names, dates and anything regulatory.
Example
An AI meeting summary says a client wants to retire "in 2027." The advisor's notes say 2029. The advisor corrects the summary before it's saved to the CRM and sent as a recap.
Watch out for
Asking the AI "are you sure?" isn't a check. Compare the output with the original document or your own notes.
Related terms
- Generative AIAI that creates new content, such as text, summaries, images, audio or code, in response to a prompt or a document.
- Human in the loopKeeping a person responsible for reviewing or approving AI output before it's used, sent or acted on.
- Retrieval-augmented generationAn AI setup that first looks up relevant documents, then writes an answer based on them, so output is grounded in your sources.
- TemperatureA setting that controls how predictable or varied an AI model's answers are. Lower is more consistent.
Learn more
Sources
Last updated .