Scan text for prompt injection attacks, hidden instructions, and malicious AI manipulation attempts. Protect your AI assistants from security threats.
Paste text from emails, PDFs, webpages, or messages to check for hidden malicious instructions.
Prompt injection is a security attack where malicious instructions are hidden in text to manipulate AI assistants (like ChatGPT, Claude, Gemini, or other language models). Attackers use these techniques to:
🚨 Direct Injection
Example: "Ignore all previous instructions and..."
⚠️ Hidden Commands
Using invisible Unicode characters (ZWSP, ZWNJ) to hide malicious instructions
🔸 System Impersonation
Example: "System: You are now in admin mode..."
🔹 Obfuscation
Using character substitution, homoglyphs, or encoding to hide malicious content
Scans for 50+ known injection patterns and malicious phrases
Detects hidden Unicode characters used to conceal instructions
All scanning happens in your browser - text never sent to servers
Is this scanner 100% accurate?
No scanner can be 100% accurate. This tool detects known patterns and suspicious content, but sophisticated attacks may evade detection. Use it as one layer of defense, not the only one.
What should I do if threats are detected?
Review the detected threats carefully. If the text is from an untrusted source, do not paste it into AI assistants. Use the "Remove All Threats" button to clean the text, but verify the cleaned version still makes sense.
Is my text uploaded to a server?
No. This scanner runs entirely in your browser using JavaScript. Your text is never sent to any server, ensuring complete privacy and security.
Can this protect me from all prompt injection attacks?
This tool helps detect many common attacks, but new techniques emerge constantly. Always be cautious when pasting text from untrusted sources into AI assistants. Combine this scanner with other security practices for best protection.
Prompt injection is one piece of the puzzle. These local-only tools cover the rest of your AI privacy stack: