Built a browser extension that stops you from accidentally sending PII to ChatGPT/Claude/Gemini

Protect Your Privacy While Using AI Chatbots: Introducing a Browser Extension for PII Prevention

As artificial intelligence tools like ChatGPT, Claude, and Gemini become integral to daily workflows, a common concern has emerged: the accidental sharing of sensitive data. From personal emails and phone numbers to API keys and social security numbers, users often inadvertently paste confidential information into chat interfaces, risking data leaks or privacy breaches.

To address this challenge, a developer has created a specialized browser extension designed to proactively prevent the unintentional transmission of Personally Identifiable Information (PII) to AI chat services. This solution offers a layer of security by intercepting messages before they leave your browser, scanning for sensitive content, and either redacting or blocking such data from being sent.

Key Features of the Extension

• Pre-Send Interception: The extension actively monitors messages directed toward ChatGPT, Claude, Gemini, and similar AI chat platforms, intercepting data before it reaches the provider.
• Comprehensive PII Detection: It scans for a wide array of PII, including emails, Social Security numbers, credit card details, phone numbers, and more.
• Automated Redaction or Blocking: Upon detecting sensitive information, the extension can either redact the PII within your message or prevent it from being sent altogether, giving users control over their data.
• Local Processing for Privacy: All scans and redactions happen locally on your machine, ensuring that no sensitive data leaves your device unless you opt for logging via the optional dashboard.
• Open Source and Community-Driven: Built with transparency in mind, the project is open source under the MIT license, inviting community contributions and enhancements.

Practical Use and Future Plans

This tool is particularly valuable for professionals in regulated industries or anyone concerned about privacy when interacting with AI services. Its straightforward implementation provides peace of mind, taking the worry out of sharing sensitive information.

While the current version demonstrates promising capabilities, it is still a work in progress. The developer plans to integrate a WebAssembly (WASM)-based PII detection model, enabling the extension to run entirely within the browser without any backend dependencies. This enhancement will further strengthen privacy guarantees and streamline performance.

Get Involved

The project is hosted on GitHub, where interested users and contributors can review the code, report issues, or suggest new PII patterns to detect. As this is an evolving tool, feedback and community input are highly encouraged to make it more robust and comprehensive.

**Repository