— Browser extension · v1.0
A browser extension that strikes through sensitive data — emails, SSNs, credit cards, API keys — before the model ever lays eyes on it.
Hi Sarah Chen,
I've reviewed the ticket from [email protected].
Card on file ending 4242 4242 4242 4242 —
refund issued. Internal ref:
SSN 123-45-6789 · +1-415-555-0117
Best,
Daniel
of employees admit to pasting sensitive company data into AI tools without sanitizing it first.
maximum GDPR fine for unauthorized personal-data processing — including third-party AI exposure.
of pasted prompts can be made to disappear from a model's training pipeline once they're in. There's no undo.
When you paste into a supported AI chat, the text is scanned against 20+ on-device classifiers. Any match is replaced with a labelled redaction before the prompt ever leaves the browser.
Already typed something sensitive into the chat box? Press Ctrl+Shift+A. Every match in your message is redacted in place — without losing the rest of what you wrote.
Every prompt is a paper trail. Treat it like one.
All processing happens on-device. There is no server. There never will be one.
No analytics, no telemetry, no usage stats. We do not know what you redact. We do not want to.
The extension is free for individuals. No upsell. No enterprise dark patterns. No ad in your settings panel.
Every detector is documented. The redaction list is exportable. Compliance reviews are a five-minute job.
When you paste into a supported AI tool, the extension scans your text against 20+ on-device classifiers. Any match — emails, SSNs, credit cards, API keys, names, phone numbers — is replaced with a labelled placeholder before the prompt is submitted. Your original text is never sent.
Yes. All processing is local — nothing reaches our servers (we don't have any). Suitable for teams that need to enforce GDPR-aligned policies on third-party AI usage without blocking those tools entirely.
Yes. Placeholders are labelled ([EMAIL_REMOVED], [SSN_REMOVED], [NAME_REMOVED]) so the model still understands the structure and intent — it just doesn't see the sensitive values.
Yes. Custom regex rules let you cover project codenames, internal URLs, customer IDs or anything else you'd rather not leak.
ChatGPT, Claude, Perplexity, Gemini, Copilot, POE, You.com, DeepSeek, HuggingFace, Mistral, Pi.ai, Character.AI, Meta AI, Kimi, Groq — and 18+ more.
Free. No account, no upsell, no telemetry. We may eventually offer a paid team tier for centrally-managed policies, but the core extension stays free.
— Get the extension
One-click install. No account. No tracking. Free for individuals — forever.