Skip to content
Get in touch

Needs review · content version 2

Review Hallucination

Your draft is bound to this exact source hash. If the term changes, the Review Desk will use a new storage key instead of silently attaching old decisions.

Definition

A hallucination is AI-generated content that presents false, fabricated, or unsupported information as though it were a valid response.
Decision for Definition

Plain-Language Explanation

A generative model can produce names, citations, events, explanations, or details that fit the language pattern of an answer without being supported by reliable evidence.
Decision for Plain-Language Explanation

Why It Matters

Fluent presentation can make unsupported claims easy to accept. Consequences grow when people use them in legal, medical, financial, operational, or public decisions.
Decision for Why It Matters

Example

A model supplies a realistic-looking court citation that does not exist when asked to support a legal argument.
Decision for Example

Common Misconception

A hallucination is not necessarily a deliberate lie or a rare malfunction. It is a possible result of how generative models produce outputs.
Decision for Common Misconception

Remember It This Way

Plausible language is not evidence.
Decision for Remember It This Way

Related Terms

* [[confidence-vs-correctness|Confidence vs. Correctness]] * [[verification|Verification]] * [[trust-borrowing|Trust Borrowing]] * [[generative-ai|Generative AI]]
Decision for Related Terms

Sources

* [NIST, *Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile* (NIST AI 600-1)](https://doi.org/10.6028/NIST.AI.600-1) — defines and addresses the closely related risk term “confabulation” for confidently presented false or erroneous generated content.
Decision for Sources

Learn More

* Apply [[verification|Verification]] before relying on generated claims.
Decision for Learn More
Return to term

Not saved yet.