SafePrompts — Spot sensitive data before sending prompts to AI

Detect emails, credentials, personal data, and other sensitive details locally in your browser — before you send. No prompt uploads.

SafePrompts helps you spot risky prompts before you send them to AI. It runs locally in your browser and warns when a prompt may contain sensitive or risky information. What SafePrompts checks for: • Secrets such as API keys, tokens, and passwords • Personal data like emails, phone numbers, IDs, and medical identifiers • Financial and payment information • Confidential business or internal information • Over-delegation to AI where human review may be needed If a risk is detected, SafePrompts explains the issue and suggests a safer way to phrase the prompt. Key features: • Local analysis in the browser • No prompt data sent to SafePrompts servers • Clear, education-focused warnings • One-click safer prompt suggestions • Works on common AI chat interfaces SafePrompts is designed as an education-first tool to help users develop safer prompting habits when working with AI. • Improved detection for internal and confidential documents • Better identification of spreadsheet and dataset pastes • Reduced noise from basic email and phone detection • Improved risk prioritization This extension provides guidance only and does not replace organizational policies or professional advice. New Release note: Improved sensitive data detection and reduced false positives. Updated messaging to clearly focus on spotting sensitive information before sending prompts to AI. Refined UI copy for clarity and consistency. All checks remain local and on-device.