Rebuff AI Description
Store the embeddings from previous attacks in a database of vectors to recognize and prevent them in the future. Use a dedicated LLM for analyzing incoming prompts to identify potential attacks. Add canary tokens in prompts to detect leakages. This allows the framework to store embedded embeddings of the incoming prompt into the vector database to prevent future attacks. Filter out malicious input before it reaches LLM.
Integrations
Company Details
Company:
Rebuff AI
Website:
www.rebuff.ai/
Recommended Products
Product Details
Platforms
SaaS
Type of Training
Documentation
Customer Support
Online