Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn to implement Zero Trust security principles specifically designed for multi-agent AI systems in this conference talk by cybersecurity experts Surendra Narang and Venkata Gopi Kolla. Discover why traditional trust models fail in multi-agent environments and explore comprehensive strategies for securing AI agent interactions through restrictive access controls, role-based permissions, and continuous monitoring. Understand how to define clear roles for each agent from system inception, implement least-privilege access principles that grant agents only necessary tools and information, and establish robust monitoring systems to detect anomalous behavior or agents operating outside their designated parameters. Examine the critical balance between maintaining system security and preserving operational speed, ensuring that security measures enhance rather than hinder agent performance. Gain insights into evaluating security risks specific to multi-agent systems including prompt injection attacks, rogue agent behavior, data leakage vulnerabilities, and adversarial manipulation techniques. Explore practical approaches for integrating trust mechanisms, enforcing secure communication channels between agents, and deploying strict access controls within decentralized agent environments. Benefit from the speakers' extensive experience in cybersecurity, cloud infrastructure, and edge computing as they share real-world strategies for securing AI-powered systems at enterprise scale.
