Prompt Guard Privacy Policy
Effective date: April 20, 2026
Prompt Guard for AI Chat reviews prompts on supported AI chat websites before they are sent. It uses local checks in the browser and, depending on configuration, may also send review requests to a configured backend service or AI review provider.
Information processed
Prompt Guard may process:
- prompt text entered on supported AI chat websites
- supported site information, page URL, and prompt-review metadata
- user activity related to the extension feature, such as send attempts, review actions, review results, and timestamps
- extension settings, device name, local prompt log entries, and a hashed settings password stored locally in the browser profile
- an optional saved NVIDIA API key, if the user provides one
- sensitive content entered into prompts, only for the purpose of detecting and blocking risky submissions
How information is used
Prompt Guard uses this information to:
- review prompts before they are sent
- warn or block prompts that appear risky, unsafe, sensitive, or low quality
- store settings and keep the extension working across sessions
- display local prompt history
- provide optional backend review results and shared dashboard features when a backend is configured
Sharing
Prompt Guard does not sell user data.
Information may be shared only as needed to provide the extension’s feature, including:
- a configured Prompt Guard backend
- a configured AI review provider such as NVIDIA
- a local model endpoint such as Ollama, if the user chooses it
- legal or security disclosures when required by law or reasonably necessary to investigate abuse or protect users
If a backend server is used, the operator of that backend may be able to view logged prompts and review results.
Storage and retention
Extension settings, the hashed settings password, device identity, optional saved API key, and the local prompt log are stored in the browser profile using Chrome extension storage.
If a backend server is used, retention depends on that server’s configuration. In the current Prompt Guard backend, server log entries remain until the server operator clears them.
Data security
Prompt Guard stores the settings password in hashed form locally. Other locally saved settings are stored in the browser’s extension storage area. If a remote backend or AI provider is configured, data is transmitted to those services according to the selected configuration.
Your choices
You can:
- turn Prompt Guard off in extension settings
- clear the local prompt log
- change the selected review provider or backend endpoint
- remove the extension at any time
Contact
For privacy questions, use the contact email listed in the Chrome Web Store developer account or store listing for Prompt Guard.