Proofreading Text and Flagging Unsupported Claims with AI
Use a structured prompt to edit text, find language errors and flag statements that need a source, with a workflow that avoids accidental misinformation.
Editing and fact-checking are different jobs. Ask for them separately so the model does not quietly rewrite your meaning.
Key points
- First pass: grammar and clarity only.
- Second pass: list claims that need a source.
- Keep the author's voice unless told otherwise.
Try this prompt
You are a careful editor. Correct grammar and clarity in the text below without changing its meaning or tone. Then list every factual claim that should be checked, with the reason. Text: [text]
Check flagged claims against trusted sources yourself; the model cannot verify facts on its own.