Hello everyone, this is my first time posting on ProductHunt. Excited to be part of this community.

The problem I am solving

You've heard it a dozen times: "Do not upload any sensitive company data to ChatGPT."

Well, people do paste and upload tons of sensitive information to AI tools. All the time. According to reports, on average someone pastes sensitive corporate data to these AI tools almost 4 times a day. This leads to violations in GDPR / HIPAA / SOC2 depending on the context of the information (eg: a medical professional uploading patient records to ChatGPT to get a diagnostic is violating HIPAA).

However, it is difficult to change user behavior. They want to keep using the superpowers of AI without any additional overhead or effort.

The solution

I have created a Chrome extension called Blankit, which redacts sensitive PII with two philosophies:

Zero trust: All data is processed on your browser. No data (raw or redacted) ever goes beyond your device. No network calls. Not even analytics.
Zero friction: After downloading, we do not expect nor want user behavior to change. Users will still interact with their AI tools as always. Blankit works in the background, protecting the user from PII leaks.

This extension is free and is available to try out here.

Feature overview

So what makes Blankit so good? Here are the key features:

🤖 Automatic redaction: Any outgoing messages to AI tools (whether it be plain-text or document files) are intercepted and the extensions scrubs PII in transit. The AI tool then gets to see only code names (eg: Alex -> [NAME_1], Stacy -> [NAME_2]). These codes persist, so you can ask follow up questions about Alex and the extension will auto-convert it to [NAME_1]. There is no additional step the user has to do.
There is also an easy toggle to put Blankit to sleep in case you want to have a conversation that does consume PII.
👁️ In-page unredact toggle: There is a button in the LLM page that lets you toggle between redacted placeholders and original values without leaving the conversation, for ease of use.
🛠️ Granular control: Toggle 17 individual detection categories (emails, phones, SSN, credit cards, addresses, names, dates, medical records, IP addresses, passport numbers, driver’s licenses, tax IDs, bank accounts, MAC addresses, URLs, credentials, UUIDs)
🔑 Custom word redaction: Define your own words to redact. Maybe you have a secret internal codeword (ex: "Project Titan") that you want to be redacted as well. Now you can do it! Plus, you can import and export this list of codewords and send it to your friends/colleagues so your group is protected with minimal effort.
📄 Audit reports: Export a PDF compliance report from the audit log, including executive summary, category breakdowns, platform stats, and recent activity. Good for trust verification and if an audit/compliance request comes knocking on your door.
💻 100% local: Zero data exfiltration. No external servers. No analytics. No telemetry. Nothing leaves your browser.
💣 Nuke button: Clear all your session and redaction information forever.

Currently, we support ChatGPT, Gemini, and Claude. I am planning to increase the support coverage to Grok and Mistral as well.

Please try it out and let me know what you think! Just install the extension, go to your AI tool of choice, and either send a plain message or upload a document with PII and see the magic work.