Privacy and your data
Sanitized AI is built to protect data, including yours. Your organization sees that something sensitive was caught, not what you wrote.
What’s recorded when something is flagged
When a prompt or a file is flagged, Sanitized AI records an event with:
- Who it was, and when.
- Which AI tool it was headed to.
- Which guardrails fired, and what kind of data they found, such as Email or Credit card.
- How severe the data is.
- What you did next: sanitized it, edited it, sent it anyway, closed the warning, or reported a false positive.
- For a file, the file’s name. If your organization masks file names, only
the file’s type is recorded, such as
•••.docx.
What isn’t recorded
- The text of your prompts.
- The AI tool’s replies.
- The contents of files you attach.
- Anything you do on other websites. The extension only works on AI websites, and only checks the ones your organization protects.
To check a prompt or a file, the extension sends it securely to Sanitized AI. It’s checked, and it isn’t stored.
Other records
- Sign-in and version: Sanitized AI records when you sign in, and which version of the extension you use.
- IT-managed computers: if your IT team installed the extension, it can record flagged activity before you sign in and after you sign out, without showing warnings.
Usage records
Your organization can also choose to record AI usage, to see how much AI tools are used. A usage record says that you sent a prompt, on which AI tool, when, and whether anything was flagged. It never includes what you wrote.
When you choose to share text
Some text is shared only when you decide to share it:
- False-positive reports. When you report a flag as wrong, the items you select, with a little text on each side of them, and your comment go to your admins and to the Sanitized AI team, who review reports to improve detection. Your full prompt doesn’t.
- Copy Details for Admin. This copies a summary of a blocked prompt: the website, the time, the rule, and the type of data. It never includes your prompt. You decide where to paste it.
Real details on your screen
If you use See real details in AI replies, your real details are kept in your browser tab only, so the extension can show them in the AI’s reply. The AI tool never receives them. They’re cleared when you close the tab, and after 24 hours at most.
Who can see what
The following table shows what each person can see.
| Who | What they see |
|---|---|
| You | Your own statistics in the extension. With dashboard access, your events on My Activity. |
| Your manager | Events for the people who report to them, including you, if they have a team dashboard. |
| Your organization’s admins | Events for everyone in your organization, and false-positive reports. |
| The Sanitized AI team | False-positive reports that you choose to send and, to operate the service, event records (who, when, which AI tool, which guardrail, and the outcome). Never your prompt text, the AI’s replies, or file names. |
| Nobody | Your full prompts, or the AI’s replies. |
More information
For security reviews, data processing questions, and compliance documentation, see the Sanitized AI Trust Center, or ask your account contact.