llm security_error ai_generated true

Prompt injection attempt detected in user input

ID: llm/prompt-injection-detected

Also available as: JSON · Markdown
72%Fix Rate
78%Confidence
3Evidence
2023-01-01First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
any active — — —

Root Cause

User attempting to override system prompt via injection.

generic

Workarounds

  1. 85% success Use separate system/user message roles and input sanitization
  2. 88% success Implement output validation and guardrails post-generation

Dead Ends

Common approaches that don't work:

  1. Filter specific known injection phrases 80% fail

    Easily bypassed with paraphrasing

  2. Trust LLM to self-police injection 82% fail

    Model can be manipulated