High Tech Institute and Cydrill organized a 45 minutes session on October 6, 2020 that gives you a thorough overview of how ML applications can be hacked, and what you can do about it.

This recorded webinar is an excerpt from the brand new face to face or online course on machine learning security that High Tech Institute and its partner for software security Cydrill are launching.

In this webinar, security expert Balázs Kiss will teach you:

  • About the cat and mouse game of software security;
  • Why machine learning security is important, and why it is difficult;
  • About the many ways the bad guys can compromise your ML systems;
  • Some real-world attacks on machine learning systems and how to defend against them;
  • How Cydrill courses can raise your paranoia to a healthy level and make your machine learning systems more robust and secure.



  • What makes machine learning a valuable target?
  • Threats from the real world:
    – Some real-world abuse examples
    – Dealing with AI/ML threats in software security

Machine Learning Security

  • Adversarial ML examples
    – Poisoning and evasion attacks
    – Demo – ML evasion attack
    – Case studies
  • The ML supply chain
    – TensorFlow security issues and vulnerabilities

Learning how not to code

Conclusion, Q&A

Presenter: Balázs Kiss

Balázs has been working in the software security field for more than 13 years as a security evaluator, researcher, and mentor. Recently, he has focused on helping developers learn how typical vulnerabilities are introduced during software development and how to stop these problems at the source. To date, he has taught more than 60 training courses worldwide.