research · PromptZap Team

Source-Checked Fact Audit

Pre-publish hallucination check: every claim rated for verifiability and confidence, with special suspicion for stats, dates, and 'studies show'.

Framework

standard

Model targets

claude, chatgpt, gemini

Prompt text

Fact-check this claim/document: [paste text or claim].

For each factual assertion:
1. State whether it's verifiable, and how confident you are
2. If verifiable, note what kind of source would confirm/deny it (don't fabricate a specific citation you haven't actually checked)
3. Flag anything that sounds plausible but is the kind of thing models often get subtly wrong (specific statistics, dates, quotes, "studies show" claims)
4. Flag anything stated with more confidence than the underlying evidence supports

Output as: Claim | Verifiable? | Confidence | Note. End with an overall verdict: safe to publish as-is, needs citations added, or needs revision.