Reviewer guidance
False positives in AI detection
Understand why AI detectors can flag human writing and how to review uncertain reports without overclaiming.
Run a signal check
No detector model loads until you press Check.
Current support
False positives are a known detector risk, especially for formal writing, non-native English, short samples, and heavily templated content.
What this checks
- AI Detect keeps classifier output as a signal, not a verdict.
- Reports include limitations and recommended next steps to avoid overclaiming.
- Reviewer workflows should compare source drafts, edit history, citations, and provenance metadata.
What this does not prove
- A high AI text signal can still be wrong.
- Human writing styles vary widely across domains and languages.
- No detector result should be the only basis for academic, employment, or legal action.
FAQ
- Can this prove whether content was written by AI?
No. AI Detect reports signals such as provenance metadata, watermark support status, and statistical text risk. It does not claim legal or authorship proof.
- Does text leave my browser during a local check?
Browser-default checks run local text features and provenance checks after you press Check. Server enhanced checks are separate and will be labelled before use.
- Why can a result be unknown?
Many providers do not expose public watermark detectors, and metadata can be absent or stripped. Unknown is a valid result when no reliable public signal exists.
- What should I do with a high signal result?
Treat it as a prompt for review. Ask for drafts, source notes, metadata, and context before making a decision.
Workflow upgrade
Need repeatable reviews?
Move from one-off checks to saved reports, batch review, and API-ready signal records when the workflow needs it.