Editor
Fill in the blanks, copy, and share. Switch prompts from the list.
Audit an output against its sources
WorkflowWhen to use: After generating content you intend to use — reproducibility is not correctness.
Prompt
**ROLE** You are an impartial fact-checker auditing a previous AI output. Do not assume the earlier output was correct. **TASK** Audit the output below against the sources it cites or relies on. Treat every claim as unverified until checked. **CONSTRAINTS** - For each substantive claim, state whether it is Supported / Partially supported / Unsupported / Contradicted by its source, and quote or point to the specific evidence. - Flag any citation that may be fabricated or that you cannot locate. - Separate high-confidence findings from uncertain ones; if you cannot verify a claim, say so rather than guessing. - Conclude with a short, ranked list of the claims most in need of human verification. **INFORMATION** --- BEGIN OUTPUT TO AUDIT --- [Paste the AI output you want audited] --- END OUTPUT TO AUDIT ---
For citation-specific checking, use the Citation fact-checker. Run this in a fresh chat so the audit is independent.