Panoptes

Panoptes: detect PHI, PII, credentials, and financial data in AI chat prompts before you send them. Local, on-device scanning.

Panoptes is a privacy watchdog for AI chat. It scans what you type into AI assistants and warns you the moment it spots sensitive data such as protected health information, personal information, credentials, or financial data, so you can catch a mistake before it leaves your browser. HOW IT PROTECTS YOU - Real-time scanning of the text you enter into supported AI chat services. - A clear warning the instant sensitive data is detected, before you send. - Optional blocking: turn on "Block on detection" and Panoptes stops the submission when it finds sensitive data, instead of only warning. PRIVATE BY DESIGN - Detection runs entirely on your device. Panoptes uses no cloud service and makes no network request to analyze your text. - No analytics, no telemetry, no call-home. Nothing is sent to the developer, ever. - Out of the box, nothing leaves your machine. Alerting to your own webhook or monitoring server is optional, off by default, and sends only redacted metadata, never your raw prompt text. SUPPORTED AI SERVICES Panoptes works across the major AI chat assistants. You choose which services to monitor from the options page, and each can be turned on or off independently. IMAGE SCANNING (BETA) Panoptes can also read text out of images you attach to a prompt, using OCR that runs on your device, and check that text for sensitive data. This feature is off by default and marked Beta: OCR misreads text, so it can both miss real data and flag data that is not there. Its accuracy has not been fully measured yet, so it only warns and never blocks. English text only, on supported AI chat services. WHO IT IS FOR Anyone who works with sensitive data and uses AI chat, whether in healthcare, finance, or legal, or anyone who would rather not paste a client record or an API key into a chat box by accident. Panoptes is independently built and maintained by Ironwright.