Prompt Guard
Named by 3 kept items in this workspace.
2 items name it under tools and 1 under affected products.
Search every workspace for Prompt Guard. It matches the name as written, so it can find other things that share it.
The items that name it 3
protects LLM powered applications from malicious prompts, including prompt injection and jailbreaking attacks
Input classifier used to flag known prompt injection patterns; catches ~30-60% of attacks depending on tuning.
Meta, 2024; DeBERTa-based filter defense evaluated
All technologies in this workspace.