Detectors for AI runtime control
Guardion provides multiple detectors to protect your AI systems from various threats and ensure safe, reliable outputs.Available Detectors
Prompt Defense
Protect against prompt injections, jailbreaks, and other adversarial attacks with our advanced Modern Guard detector.
Modern Guard
Our flagship detector that uses representation engineering to identify and block sophisticated prompt attacks.
Coming Soon
Content Moderation
Filter harmful, inappropriate, or unsafe content from both inputs and outputs.
Hallucination Detection
Identify uncertainty in AI responses using representation engineering techniques.
PII Detection
Automatically detect and redact personally identifiable information in prompts and responses.
Output Handling
Safely manage code execution and other potentially risky outputs.
More Guardrails
Additional safety features coming to the Guardion platform.
Detector Configuration
Each detector can be configured with custom thresholds and policies through the Guardion dashboard or API.How do detectors work?
How do detectors work?
Guardion’s detectors use advanced machine learning models to analyze inputs and outputs, identifying patterns that match known attack vectors or unsafe content.
Can I customize detection thresholds?
Can I customize detection thresholds?
Yes, each detector allows for custom confidence thresholds to balance security with usability for your specific use case.
How do I implement guardrails?
How do I implement guardrails?
Guardrails can be implemented via our API or SDK. Check our Quickstart guide for implementation details.