Skip to content
Sanitized AI Help Center

Quickstart

Your organization uses Sanitized AI to keep sensitive data out of AI tools. Here’s how to get protected. There’s no password to create.

  1. Check your welcome email.

    When your admin activates your account, you get an email titled Welcome to Sanitized AI: your account is ready. It lists the steps for your account. If you don’t have it, check your spam folder or ask your admin.

    If your organization creates accounts through its company sign-in, you might not get a welcome email. Go straight to the next step.

  2. Install the browser extension.

    Install it from your browser’s store: Chrome Web Store (also for Brave), Microsoft Edge Add-ons, or Firefox Add-ons.

    If your IT team manages your computer, the extension may already be installed. Your welcome email says so. For details, see Install and sign in.

  3. Sign in to the extension.

    Click the Sanitized AI icon in your browser’s toolbar, enter your work email, and click Continue. Then click Email me a sign-in link and open the link from your inbox in the same browser, or sign in with your company account. A You’re protected screen confirms that you’re signed in.

  4. Use AI tools as usual.

    Sanitized AI checks each prompt before it’s sent. You only hear from it when a prompt contains something sensitive.

What you’ll see when a prompt is flagged

A Security Check appears in the corner of the page. It shows what was found and why it matters. You can:

  • Click Sanitize Prompt to preview your prompt with the sensitive parts replaced by safe stand-ins, and then click Insert Sanitized to send it.
  • Click Edit Prompt to reword it yourself.
  • Click Send Anyway, if your organization allows it for that data. Your organization is told that you sent it anyway.

Some data can never be sent in a prompt. In that case the warning reads Blocked by Policy and has no Send Anyway button. For every option, see When a prompt is flagged.

Check that it’s working

Type a prompt with a made-up name and phone number, such as Call Jane Doe at (514) 555-0134, into an AI tool that your organization protects, such as ChatGPT. A Security Check should appear before the prompt is sent.

If nothing appears, see Troubleshooting. Some organizations start in a monitoring mode where nothing is shown to you; your admin can tell you.