Stop your team from pasting customer data into ChatGPT.
Cairn Guard catches emails, credit cards, SSNs, API keys, and other sensitive data in your prompt before it's sent to ChatGPT, Claude, Gemini, or Copilot. Everything runs in the browser. Nothing ever leaves the device.
What it looks like in flight
A quick warning, not a roadblock. Your team keeps moving — they just don't accidentally send the wrong thing.
ja••••••••••••••••••••••om42•••••••••••••4255•••••56(4••••••••••34What it catches
Detection runs entirely in your browser — regex, validators, and contextual heuristics. No model calls, no per-request cost.
Credentials & secrets
AWS, GitHub, Slack, OpenAI, Anthropic tokens. JWTs. Private-key blocks pasted from terminals.
Financial
Credit-card numbers (Luhn-validated, never false-positives on order IDs). IBANs.
Government IDs
US Social Security Numbers (format-validated). Passport numbers (with context cue).
Personal info
Emails, phone numbers (US + international), street addresses, ZIP codes, dates of birth.
Network & hosts
IPv4 addresses, MAC addresses, internal hostnames.
Custom rules (team)
Org-specific patterns: customer IDs, internal product codes, ticket references, anything else you don't want in a chat box.
Pricing
Free for individuals. Real pricing for teams that need rollout, audit, and admin.
- Unlimited detections
- Custom regex rules
- Priority support
- License works on all your devices
- Everything in Solo
- Shared organization rules
- Block-event audit log (no content stored)
- Admin invite & rollout
- Email support, 1 business day
Need SOC2 docs, custom DLP integrations, or 100+ seats? Email us.
Questions
Does my prompt ever leave my browser?
No. Detection happens locally — all the regex and validators run in the extension's content script. We don't have a server that sees your prompts, even on the team plan. The only thing the optional team audit log records is metadata (which categories were detected, when, and on which host) — never the matched content itself.
How is this different from PrivacyScrubber, Privacy Shield, etc.?
Three things. (1) Pricing — $5 / seat for teams, vs. $99 / month flat or enterprise-only contracts elsewhere. (2) Real team rollout — admin console, audit log, shared rules — without an enterprise sales cycle. (3) Open-source detection rules so your security team can audit what we're checking for.
Can I add my own organization-specific patterns?
Yes — on the Solo and Team plans you can add custom regex rules. On the Team plan, those rules sync across all seats so a security admin can roll out org-specific detections (customer-ID formats, internal product codes, etc.) once.
What happens when a detection fires?
A modal appears showing what was detected and where. The user can Redact and send (placeholders replace matches before submit), Send anyway (the team audit log records the override), or Cancel (the prompt stays in the textbox unchanged).
Which sites does it support?
ChatGPT (chatgpt.com and chat.openai.com), Claude (claude.ai), Gemini (gemini.google.com), and Microsoft Copilot (copilot.microsoft.com). More on the way — request a site and we'll usually add it within a day.
Browser support?
Chrome and Edge today (both Chromium-based). Firefox MV3 port is on the roadmap.
Refunds?
30 days, no questions. Email and we'll process it.
Add Cairn Guard to your team's browser today.
Free for individuals. $5 / seat / month for teams. No data leaves your device.
Install free from the Chrome Web Store