Manual red teaming assessments are time consuming. Both scaling and repeating the process for all your applications and use cases.

We have 3 different automated approaches for LLM vulnerabilities. Here we will be focusing on a single vulnerability named “Prompt Injections” using all the approaches.

  1. Manually Testing for Prompt Injection
  2. Using a Library of Prompts Technique
  3. Using Giskard LLM Scan to Identify Prompt Injection

DEMO LLM APPLICATION:-

TESTING LINK : https://s172-29-61-157p8888.lab-aws-production.deeplearning.ai/notebooks/L3/L3_Red_teaming_at_scale.ipynb

GISKARD LLM SCAN:-

image.png