Everything you need to know about using PromptSafe to protect your sensitive data when using AI tools.
PromptSafe is a data anonymization tool that lets you use AI tools like ChatGPT and Claude on real work — client emails, contracts, HR files, legal documents — without exposing anyone's private information. It automatically detects sensitive data and replaces it with readable placeholders like [PERSON_1] or [COMPANY_2] before you share anything with AI. After you get your AI response, you can restore the original data using your Restoration Key.
Anyone who works with other people's data and wants to use AI to speed up their work. That includes paralegals handling discovery documents, HR professionals drafting PIPs from employee records, consultants analyzing client financials, recruiters screening resumes, accountants reviewing tax docs, and small business owners processing customer information. If you've ever thought "I'd love to use ChatGPT for this, but I can't because of the client data" — PromptSafe is for you.
Manual redaction is slow, error-prone, and doesn't scale. A single contract might contain dozens of names, emails, and addresses scattered across pages. PromptSafe detects all of them automatically, replaces them consistently (every instance of "Sarah Chen" becomes [PERSON_1] everywhere), and gives you a Restoration Key so you can reverse it afterward. What takes 15–20 minutes manually takes seconds with PromptSafe.
When you upload a PDF, Word doc, or any file directly to an AI tool, the entire file — including every name, email, ID number, and address inside — is sent to their servers. Most people don't check what's inside a file before uploading. That's the danger: you may be sharing dozens of people's personal data without realising it. PromptSafe gives you two safer options: redact the file first and upload the clean version, or bypass the upload entirely by extracting the text, anonymizing it, and pasting it into the AI chat as safe text instead.
No. PromptSafe runs 100% in your browser. Your documents and text never leave your device — nothing is sent to any server. You can verify this yourself: open your browser's DevTools (F12), go to the Network tab, and watch while you scan a document. You'll see zero outbound data requests. This isn't just a policy — it's the architecture.
By default, text you paste or files you upload to AI tools may be stored on their servers or used for model training. PromptSafe solves this by replacing all sensitive information with safe placeholders before you share anything. The AI gets all the context it needs to do useful work — without the real names, emails, or personal details.
PromptSafe Text works with plain text — paste or type content on the website, review the detected items, anonymize, and copy the safe text into any AI tool. You can also upload files (PDF, Word docs) and PromptSafe will extract the text for you, so you can anonymize it and paste it into AI as safe text — bypassing the need to upload the original file to the AI tool at all.
PromptSafe PDF preserves the visual layout of your document. Upload a PDF, and PromptSafe applies redaction overlays directly onto the pages — you can see exactly what's being replaced, where. Download a cleanly redacted PDF that looks professional enough to share with colleagues or upload to an AI tool.
The Chrome Extension brings PromptSafe directly into ChatGPT and Claude — no tab-switching needed. Drop a document into the side panel, and the extracted text is scanned and pushed into the AI chat box with items highlighted. Review, hit sanitize, and send. It's the fastest workflow for professionals who process documents through AI daily.
Use PromptSafe PDF when the document's formatting matters — tables, letterheads, multi-column layouts — or when you need to visually verify that every sensitive item is covered before sharing. It's also much more helpful for large documents: rather than scrolling through walls of extracted text, you can visually navigate the actual pages and see exactly where each redaction sits. Also use it when you want to share the redacted document with other people (not just AI).
Use PromptSafe Text (or the Chrome Extension) when you want to bypass file uploads entirely. Instead of uploading a sensitive document to the AI, you can extract its contents through PromptSafe, anonymize the text, and paste it into the chat. The original file never touches AI servers.
The extension is fully built and works on ChatGPT and Claude. It's currently pending approval on the Chrome Web Store. Once accepted, it will also work on Edge, Brave, and Opera (all Chromium-based browsers). In the meantime, you can use the PromptSafe website to anonymize text and PDFs.
PromptSafe's automated detection catches the most common types of sensitive data — names, emails, phone numbers, dates, addresses, company names, ID numbers, and more. However, no automated tool catches 100% of sensitive items every time, especially unusual names or context-specific data. That's why PromptSafe includes tools to help you review and fill in the gaps.
After the automated scan, you can review your document and manually label anything the scanner missed. Just highlight the text and mark it as sensitive — PromptSafe will add it to the detection list and assign it a placeholder.
You can also use the Search function, which is especially useful for longer documents: type in a name or term, and PromptSafe will find every occurrence across the entire document and flag them all for redaction. Catch one, and you catch them all.
Between automated detection, manual highlighting, and search — you have full control over what gets anonymized.
When PromptSafe replaces data with placeholders like [PERSON_1] or [ORG_2], it creates a Restoration Key — a small JSON file that maps each placeholder back to the original value. Save it locally. After the AI gives you a response using the placeholders, paste that response back into PromptSafe and use the Restoration Key to swap the real names and data back in. The AI did the thinking; you get the real output.
Traditional redaction tools like Adobe Acrobat have three major limitations:
No automatic detection. You have to manually find and select every name, email, and phone number yourself — across every page. Miss one and it's exposed. PromptSafe scans the entire document automatically and flags everything it finds.
Black boxes destroy context. Traditional tools black out text permanently. The content is gone, and the document becomes hard to read or unusable for AI analysis. PromptSafe replaces sensitive data with readable placeholders like [PERSON_1], so the document still makes sense.
No "catch all occurrences." If "John Doe" appears 15 times across a 30-page document, you have to find and redact each one manually. In PromptSafe, highlight one instance and the app finds and redacts every occurrence across the entire document automatically.
Plus, PromptSafe's redaction is reversible — you're not destroying anything, just protecting it temporarily.
Yes. The free tier gives you access to 5 entity types (e.g., emails, phone numbers, and names) with unlimited use — no account required. For the PDF tool, you get one free trial document to experience the full workflow. Most casual users find the free tier enough. Professionals working with complex documents that contain many types of sensitive data will hit the upgrade point naturally.
Pro ($5.99/month): Unlimited entity types for text, 5 PDFs per month with unlimited entity types, and the Restoration Key.
Business ($14.99/month): Everything in Pro, plus unlimited entity types on PDFs, early access to new features, and priority support.
Annual plans save ~17%. Lifetime options are also available for both tiers.
Enterprise redaction tools typically charge $29–99+ per month, require cloud uploads or complex on-premise deployment, and produce permanent black boxes that destroy document context. PromptSafe starts at $5.99/month, runs entirely in your browser, and produces readable, reversible output. If you're an individual professional or small team — not a Fortune 500 compliance department — PromptSafe gives you what you need at a fraction of the cost.
The web app (Text and PDF) works on any modern browser — Chrome, Edge, Firefox, Safari, Brave, Opera. The Chrome Extension works on all Chromium-based browsers (Chrome, Edge, Brave, Opera) once available on the Chrome Web Store.
The web app requires an initial page load, but once loaded, all processing happens locally in your browser. The Chrome Extension works similarly — once installed, the detection engine runs entirely on your device.
OCR support for scanned PDFs is on our roadmap. Currently, PromptSafe PDF works best with text-based PDFs where the text layer is selectable. For scanned documents, you can use an OCR tool first and then bring the extracted text into PromptSafe Text.