Sanitized AI Extension 作者： Sanitized Ai

Sanitized AI is a data loss prevention tool for organizations whose staff use AI chat assistants. It checks what an employee is about to send to an AI site and applies your administrator's policy before the text or file ever leaves the browser.

How it works

When you send a prompt or attach a file on a supported AI site, the add-on sends that content to your organization's Sanitized AI service to be scanned for sensitive data such as personal information, credentials, health records and financial details. Depending on the policy your administrator set, the prompt is allowed through, you are warned and asked to confirm, or it is blocked.

Because of this, the add-on reads what you type into supported AI sites and transmits it to your organization's own Sanitized AI service for scanning. It is inactive on every other site.

Supported sites

ChatGPT, Claude, Google Gemini, Microsoft Copilot, Grok, Perplexity, DeepSeek, Mistral, Meta AI, NotebookLM, Google AI Studio, Groq, Hugging Face, Kimi, Doubao, Pi and Poe.

Requirements

A Sanitized AI work account. The add-on does nothing until you sign in. On Firefox, sign in with your email address and password. Requires Firefox 140 or later.

For administrators

The add-on can be deployed across a managed fleet through a Firefox enterprise policy, with configuration supplied via policies.json or a native manifest.