On this episode of Pentesters Chat, our team explores the distinct security vulnerabilities that arise when testing AI/ML systems compared to traditional systems.

  • Adversarial Attacks: Understand how adversarial inputs can manipulate machine learning models, and how pentesters can exploit this weakness.
  • Model Inference: Discuss techniques for reverse-engineering AI models and extracting sensitive data, including training datasets.
  • Defense Strategies: Share insights on strengthening AI/ML systems against common attack vectors and building more resilient models.

On this episode from the Sprocket Team:

Nicholas Anastasi

Willis Vandevanter

Nate Fair

Juan Pablo (JP) Gomez Postigo

Justin Wise