Drop documents right inside ChatGPT or Claude. PromptSafe detects confidential data, replaces it with smart placeholders, and injects safe text directly into the chat. Restore originals in one click.
Also works on Edge, Brave, Opera, and all Chromium browsers.
Or use the web app
Looking for PDF anonymisation? PromptSafe PDF →
Contracts, resumes, case files, financials — any document with confidential data. PromptSafe extracts the content automatically.
Names, IDs, emails and more are auto-detected. Add anything missed, remove what doesn't need hiding.
Hit "Anonymise & Inject" — safe text goes straight into the AI chat. AI responds with placeholders. Restore originals in one click.
Toggle between original and anonymised to see what AI receives
PromptSafe detects entity types — not individual words. It scans for names, NRICs, emails, phone numbers, and more across your entire document automatically.
Spotted a false positive? Click to remove it. Missed something? Highlight any text and label it — Person, Company, NRIC, or any of 18+ types. You're always in control of what gets anonymised.
The full workflow without switching tabs. Drop documents, anonymise, inject, and restore — all from a side panel.
After installing the extension, click the PromptSafe icon next to the Microphone icon on ChatGPT.
Drag in a PDF, DOCX, or image. Text is extracted instantly with names, NRICs, emails, and financials highlighted.
Review the extracted text and add or remove any sensitive items. You're always in control of what gets anonymised.
Hit "Anonymise & Inject" — safe text with [PERSON_1] placeholders is inserted directly into the chat input.
AI responds using placeholders. Restore original data by clicking the blue banner under the output (ChatGPT) or copying the output and pasting in the extension's Restore tab (Claude).