Corroba Verify · for AI
Did ChatGPT make up your sources?
AI chatbots routinely invent citations — plausible titles, real-looking DOIs, papers that were never written. Paste the references ChatGPT, Claude, or Gemini gave you and Corroba checks every one against real databases, so you never submit a hallucinated source.
Why AI invents citations
They predict, they don't retrieve
An LLM writes the most likely next words. A citation that looks right is easy to produce even when no such paper exists.
Fake DOIs look real
Models can emit a well-formed DOI that resolves to a different paper — or to nothing. A real DOI on the wrong title is one of the most common tells.
Confidence isn't evidence
AI states invented references as confidently as real ones. The only reliable check is a lookup against a real database — which is exactly what this does.
Questions
- Does ChatGPT make up references?
- Often, yes. Large language models generate plausible-sounding text rather than retrieving real records, so they routinely produce citations that don't exist — sometimes with convincing titles and even real-looking DOIs. Studies have found a large share of AI-generated citations are fabricated or contain errors.
- How do I check citations from ChatGPT or Claude?
- Paste the reference list into the box above. Corroba looks up each entry in real bibliographic databases and tells you which are verified, which point to a different work, which have a broken DOI, and which can't be found at all.
- Is the AI citation checker free?
- Yes — up to 25 references per run, no sign-up. Corroba never invents a source to fill a gap, so a “not found” means we genuinely couldn't confirm it exists.