Testing Facial recognition system security at Secure360

The founder of Adversa Alex Polyakov delivered a talk titled ‘AI Security: Why, What, How’ at the Secure360 Conference. 

The Secure360 Conference is the professional conference of choice for comprehensive security and risk management education. Collaboration and education for your entire team.

Secure360 focuses on the following key focus areas: cybersecurity, governance, risk and compliance, physical security, business continuity management, and professional development.

Facial recognition systems have their specific deep learning models, besides, the systems usually work in the physical environment and their attack surface differs from the digital one presented in academia papers. Another problem is that all examples of attacks and defensive measures were given for various models, datasets, and conditions. This can’t help you understand the real situation with facial recognition systems even if you read about 100 research papers currently published in this topic.

To be able to make proper tests we created our own attack taxonomy to test effectiveness of recent approaches to attacking facial recognition systems. Alex presented our research conducted in the real environment with various cameras and algorithms. He also showed how to protect production systems from those kinds of attacks.

Alex demonstrated how we conducted a security assessment of multiple face recognition software and hardware solutions to find the most secure configuration.

The topic was presented already, but the research is still going on. New insights will be provided, which were not published before.