AI hallucination
A confident but fabricated output from a language model — in law, a cited case that does not exist.
An AI "hallucination" is a plausible-sounding output that is not grounded in any real source. In legal work the most dangerous form is a fabricated citation: a case name, reporter cite, or quotation that looks correct but does not exist or does not say what the model claims.
Hallucinations occur because language models predict the next likely token rather than retrieving from a database. The professional risk is concrete — lawyers have been sanctioned for filing briefs with fictional AI-generated cases. The defenses are retrieval grounding and post-hoc citation validation, plus reading every cite before filing.