AI is evolving faster than our traditional security systems can keep up, and the risks are becoming increasingly psychological. In this episode, we dive deep into the hidden world of cognitive biases within AI models and why they represent the next frontier of risk management. Join host Boris Agranovich as he speaks with Avi Pilser, CEO of Ultra Deep Tech, about the critical intersection of psychology and artificial intelligence. While most people focus on social discrimination bias, Avi explains why cognitive biases like anchoring and confirmation are the true roots of AI hallucinations and manipulation.
KEY TOPICS IN THIS EPISODE
🚀 BEYOND TRADITIONAL SECURITY Discover why the old methods of cybersecurity are becoming obsolete in the age of AI agents. Avi explains the shift toward a holistic approach that treats AI as an alien intelligence with its own unique psychology.
🔍 DECODING COGNITIVE BIAS Learn how human biases are transmitted to AI through training data and reinforcement learning. We discuss how these biases lead to hallucinations and how tools like Lucid can detect 74 different biases in real time.
⚖️ NAVIGATING THE EU AI ACT We break down the implications of Article 5 and what GRC leaders must do before August 2026 to avoid prohibited manipulative practices and maintain customer trust.
📊 MEASURABLE AI CONTROLS Avi shares practical advice for risk teams and internal auditors on how to establish transparency and explainability using advanced metrics and safety shields.
If you want to be our guest or suggest a speaker, send your email to info@globalriskconsult.com with the subject line "Podcast Guest".
🌐 Join our community for even more insights:
Online Community: https://globalriskcommunity.com/
LinkedIn Page: https://www.linkedin.com/company/globalrisk-community/
LinkedIn Group: https://www.linkedin.com/groups/3701313/
Academy platform: https://globalriskacademy.com/courses/
Comments