EMPLOYEE SELF SERVICE · BROWSER EXTENSION
Using Deidentified with AI Tools like ChatGPT and Claude
You write normally — personal information is swapped for safe placeholders before your message ever leaves the browser.
Pick profile → Send → AI sees placeholders onlyThe shield in the prompt box
A small Deidentified shield appears in the AI tool’s message box. Click it to choose which de-identification profile applies — usually your organisation’s Default Organisation Policy — or open settings via the gear icon.
The shield and profile picker in ChatGPT
What happens when you send a message
As you submit, the extension briefly shows “De-identifying your text…” while it detects and replaces personal information.
De-identification happens as you send
Your message reaches the AI with placeholders — like <FIRST_NAME>, <EMAIL> or <PHONE_NUMBER> — instead of the real values, along with an automatic note asking the AI to preserve them unchanged.
The AI only ever sees placeholders
The AI’s answer keeps the placeholders
The response comes back using the same placeholders, so it contains no personal information either. Restore the real values on your side at any time — see Re-identifying AI responses.
The response preserves the placeholders
If you see “Alert: No DeID filter is applied”
The red alert next to the message box
A red alert on the shield means your text will be sent without de-identification — usually because the No DeID profile is selected or no profile is active. Click the shield, choose your organisation’s default profile (or select Refresh Profiles), and check the alert has cleared before sending anything sensitive.