AI hallucination
A response that appears plausible but contains information that is unsupported, fabricated, or inconsistent with the relevant facts or source material.
Why it matters
Grounding, deterministic controls, validation, and human escalation can reduce risk, but no single technique removes it completely.