Poking AI in the Eye:   A Practical Intro to Adversarial AI

New York Metro Joint Cyber Security Conference  
September 26^th, 2024

@ Microsoft Technology Center - 11 Times Square, NYC

Artificial intelligence (AI) systems are becoming ubiquitous in our daily lives, but how secure are they? In this presentation, I will introduce the audience to some of the techniques used by hackers and researchers to attack AI systems, such as data poisoning, model stealing, and prompt injection. I will also demonstrate some of these attacks live.

The presentation will be interactive and engaging, with a few light challenges for those audience members who want to try their hands at breaking AI (from the comfort of your mobile phone). I will also highlight some real-world case studies of successful attacks. Whether you are an AI enthusiast, a security professional, or just curious about how AI fails, this presentation will give you a practical and fun introduction to the fascinating world of adversary AI techniques, along with ample resources to get you started with practicing these techniques legally.

Key Takeaways:

• Major attack categories facing AI systems, and their potential impacts.
• Real-world examples of successful attacks against AI systems.
• Resources will be provided to allow hands-on activities after the presentation.

Check the agenda