Privacy data is automatically masked before being sent to the AI system.
PasteGuard is an open-source privacy protection tool: before you send content to ChatGPT, Claude, various AI APIs, or programming agents, it replaces sensitive information such as names, email addresses, API keys, and customer data with placeholders like [[PERSON_1]] and [[EMAIL_1]].
What the model receives are merely placeholders, but it is still able to understand the task at hand. When the response is generated, the original text will be restored and displayed to you in scenarios where this is supported. In the official demonstrations, it was shown that a single scan takes less than 100 milliseconds on average, and the system is capable of identifying names, emails, phone numbers, passwords, JWT tokens, and other information contained within multi-language texts.
It can be deployed locally, and there are also browser extensions available. It is ideal for teams that need to use AI to process customer service tickets, logs, and code but are reluctant to send real data outside their systems.