PromptFence

Warns, redacts, or blocks sensitive data before it is sent to public AI tools.

PromptFence helps prevent sensitive information from being shared with public AI tools. The extension scans prompt text locally in your browser before submission. When it detects supported patterns, PromptFence can warn you, replace matched values with clear placeholders, or block credentials and other high-risk information according to your team policy. Key features: • Detects personal, financial, healthcare, identity, and account information • Detects API keys, passwords, access tokens, private keys, database credentials, session cookies, and other secrets • Redacts matched values while preserving the rest of your prompt • Blocks high-risk credentials before submission • Supports organization-defined sensitive terms • Provides a quieter drafting workflow with dismissible alerts and a compact risk badge • Rechecks prompts before they are sent • Connects to the PromptFence dashboard for policy management and redacted audit events • Allows users to pause protection or disconnect the browser Prompt scanning and policy decisions run locally. PromptFence does not include raw prompt text in stored audit-event payloads. Connected installations may send action metadata, finding categories, a SHA-256 content hash, the AI tool domain, and a limited redacted sample to the PromptFence dashboard. Pattern-based detection may produce false positives or miss uncommon formats. Users should review each warning before deciding whether to redact or continue. PromptFence is a TensorJetAI product for individuals and teams using public AI services.