Unlocking Backdoor AI Poisoning with Dmitrijs Trizna

This BlueHat Podcast episode features Microsoft security researcher Dmitrijs Trizna discussing AI-based threat detection for Kubernetes and Linux, the challenges of building Trustworthy AI, and how attackers use backdoor poisoning and unusual input patterns to compromise models. He highlights the importance of interdisciplinary approaches to AI security, touches on the resilience of gradient-boosted decision trees, and shares insights from his Blue Hat India presentation, where interest in AI security was notably strong.