InputGuard

Scans pasted text, prompts, and supported text attachments for sensitive data before use in AI sites.

InputGuard is an AI DLP browser extension that helps prevent sensitive data from being pasted, typed, submitted, or attached to supported AI tools. It scans prompts locally in the browser before they are sent and can warn users, require redaction, require justification, or block content based on policy. InputGuard helps teams reduce accidental exposure of sensitive information in AI workflows, including: • Social Security numbers • Credit card numbers • Bank and routing numbers • Email addresses and phone numbers • API keys and secrets • Private keys • Password-like values and connection strings • Custom business-sensitive terms For individual users, InputGuard can run locally without an account. For organizations, InputGuard provides managed AI DLP policies, deployment profiles, device inventory, and audit metadata through the InputGuard admin portal. Key features: • Local prompt scanning before AI submission • Paste protection • Submit protection • Text attachment scanning • Redaction workflows • Justification workflows for managed teams • Block rules for high-risk content • Managed policy sync • Metadata-only audit logging for organizations InputGuard does not upload prompt contents for scanning. Detection runs locally in the browser. In managed deployments, audit events are limited to metadata such as detector type, action taken, site, device, and policy outcome. Supported AI sites may vary as AI websites update their editors and interfaces. InputGuard is designed for common AI chat workflows and will continue to improve coverage over time. InputGuard is operated by Sunny Gateway LLC.