Privacy gateway for shadow AI

Stop sensitive data before public AI.

PromptFence warns, redacts, or blocks secrets, PII, and regulated data inside everyday AI tools.

Create free account Get extension

Abstract privacy gateway protecting prompt data streams

Secret blocking

Blocks credentials and tokens under balanced policy.

PII redaction

Redacts emails, phones, DOBs, addresses, IDs, and health identifiers.

Team rollout

Invite-token enrollment connects browsers to the right team policy.

Product surface

One guardrail across browser prompts, team policy, and review.

PromptFence pairs a lightweight extension with a team dashboard, so owners can set policy and users can keep working without guessing what is safe to paste.

Browser extension intercepting sensitive prompt data before send

Browser extensionInline warnings on ChatGPT, Claude, Gemini, and other public AI surfaces.

Policy dashboardPresets, custom terms, allowed domains, and scanner simulation for admins.

Event reviewRisk score, action, finding types, content hash, and redacted sample for each event.

Browser rollout

Install the guardrail where prompts are written.

PromptFence extension listings are prepared for Chrome and Firefox, with team enrollment handled through the dashboard.

ChromeChrome Web Store

Prepared for Chrome Web Store distribution across Chromium-based browsers.

Download for Chrome

FirefoxFirefox Add-ons

Prepared for Mozilla Add-ons review and managed team rollout.

Listing coming soon

Team setup stays in PromptFence.

Admins create invite tokens, connect each browser to the right team, and keep policy updates synced from the dashboard.

Manage installs

Operating model

Built for teams that want control without slowing work down.

Detect

Detect before send

The extension scans prompt text locally, flags sensitive values, and shows the safest next action before data leaves the browser.

Decide

Apply team policy

Admins choose a preset, define sensitive terms, and decide whether users can send after warning or must redact first.

Review

Keep audit trails

PromptFence stores metadata, hashes, redacted samples, and decisions so teams can review usage without retaining raw prompts.

Detection coverage

Designed for the data people paste by accident.

The scanner combines structured patterns, context terms, entropy checks, validators, and team-defined phrases.

Email and phoneSSN and tax IDsPasswordsAPI keysOAuth tokensPrivate IPsMRNsWallets