We are launching soon.Get early access
PalveronPalveron
Sign inBook a demo
PALVERON AI BROWSER GUARD

Use AI without giving data away.

Your employees keep working with ChatGPT, Claude or Copilot. The Browser Guard checks every prompt before it is sent and asks first instead of blocking outright.

The dialog uses the extension's actual wording. Anyone who wants to send the original must give a reason, which is stored with the record.

How the Browser Guard works

  1. Detect

    The Browser Guard detects the use of known AI services. In those named below it also checks every prompt before it is sent.

  2. Ask instead of block

    If person names or contact details such as an address, an email address or a phone number are in the prompt, the person decides: send it masked, or send the original with a justification. In Gemini the prompt is blocked instead of masked. The Browser Guard always makes credentials unrecognizable.

  3. Decide per service

    You decide which AI services are sanctioned. On an unsanctioned service the Browser Guard shows a warning.

  4. Prove

    Every checked prompt is stored as a record with an integrity hash, its content encrypted. Within your team, only owners and admins can see the original, and only if your project allows it.

AI SERVICES IN WHICH EVERY PROMPT IS CHECKED

  • ChatGPT
  • Claude
  • Gemini
  • Copilot
  • Perplexity
  • DeepSeek
  • Grok
  • Meta AI
  • Mistral Le Chat

Beyond these, it detects the use of further known AI services.

FOR DATA PROTECTION AND WORKS COUNCILS

What is recorded and what is not

Recorded

  • Which AI service was opened and when
  • Checked prompts as an encrypted record with an integrity hash
  • The decision: passed, sent masked, or sent in the original with a justification

Not recorded

  • Names of employees: the Browser Guard sends neither the name nor the email address of the person using it
  • Other websites: only AI services and addresses an AI service forwards to are reported
  • Prompts in browsers without the Browser Guard

What is available today

  • Chrome, Edge and other Chromium browsers
  • A check of every prompt in the AI services named above
  • Approval status per AI service, with a warning on unsanctioned services
  • A setting per project to always mask company names and places

What is planned

  • Installation from the Chrome Web Store
  • Distribution through your device management, with a separate key for each device
  • Firefox and Safari, in the first half of 2027
  • Redirection from an unsanctioned to a sanctioned AI service

Get to know the Browser Guard

In a demo you see what your employees experience and what arrives in your Palveron account.

To the platform page: AI Browser Guard
Book a demo