LLMMedium
ChatGPT Provided Non-Existent Citations and Links when Prompted by Users
- Occurred
- Nov 30, 2022
- Published
- Mar 2, 2023
- Source
- AI Incident Database
Summary
When prompted about providing references, ChatGPT was reportedly generating non-existent but convincing-looking citations and links, which is also known as "hallucination".
Developers
- OpenAI
Deployers
- OpenAI
Harmed parties
- ChatGPT users
Tags
- variant:approved
Classification
Category confidence: 95%
Severity confidence: 90%
Rationale: ChatGPT hallucinating citations is a known LLM failure mode causing misinformation without physical harm.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8