AI Privacy Preflight
Review text for common credential and personal-data patterns, then get a redaction checklist before using a public AI service.
Use the tool ↓- Identify the material
- Review detected categories
- Redact and check consent
Your answer will appear here
Set the inputs, then select “Review privacy signals”. You will see the result, its assumptions and the next useful step.
- Identify the material
- Review detected categories
- Redact and check consent
Method and worked example
Local pattern checks flag common credentials, private-key blocks, email addresses, phone-like identifiers and sensitive-topic terms. The result reports category counts and actions, never a certification that data is safe.
What this result can tell you
Pattern checks can miss names, unusual credentials and contextual disclosures, and can flag harmless numbers. A clear pattern scan is not permission to upload the text.
Learn safe prompting →Method version 3.16.0
Common questions
Is my pasted text uploaded or saved?
The check runs in memory in your browser. It makes no model request, creates no account record and clears when you refresh.
Can it certify privacy or legal compliance?
No. Check ownership, consent, provider retention and your organisation's rules yourself. Redact identities and confidential facts even when a pattern was not detected.
References and next checks
Documentation checked 2026-09-05. Provider capabilities can change; use the linked documentation for the exact model or plan.