HELP CENTER
How to use the AI prompt scanner
Use the scanner as an extra review step to look for selected sensitive patterns before sharing text with an AI service.
1. Paste safe sample text
Paste text into the scanner and choose which supported personal-data categories to check. For testing, use fictional examples rather than real passwords, API keys, private prompts, or customer information.
2. Scan and review findings
Select Scan text and review each masked finding. Detection may include common API-key or credential formats, private-key blocks, email addresses, phone-like numbers, payment-card-like numbers, and selected identifiers, depending on the patterns in the current version.
3. Create and inspect a redacted copy
The redacted copy replaces detected matches with masking text. Review the full result before using it: surrounding context may still reveal sensitive information, and undetected data may remain.
Does the website upload my prompt?
The browser-based demo scanner runs its scan locally in your browser and does not send the text you paste to a server for analysis. Avoid entering real secrets into any website unless you have independently verified its behavior and privacy terms.
Why did it miss something or flag normal text?
Pattern matching is not perfect. A secret may use an unfamiliar format, be split across text, or look unlike known patterns. Ordinary text can also resemble a credential or identifier. A clean scan is not proof that text is safe.
Does it protect every AI website?
No. The website scanner checks only text you paste into it. The separate browser extension supports selected AI websites and may not cover every input, attachment, browser, or future site change.
Is this a replacement for company security controls?
No. Teams should still use approved AI policies, access controls, secret-management practices, and appropriate data-loss-prevention tools. Treat this as an additional review aid, not a security guarantee.
Related guidance
For practical steps beyond the scanner, read our guide to protecting sensitive data in AI prompts and our AI data protection guidance for small teams.
Still need help?
Use the contact and issue-reporting page. Share only sanitized examples. Never post actual keys, credentials, private prompts, or customer information.