AI Agents Pose New Security Challenges for Defenders
Palo Alto Networks' Kevin Kin discusses the growing security risks posed by AI agents and the difficulty in distinguishing their behavior from users.
By SecurityEditor • August 8, 2025
The rapid evolution of AI-powered security tools has introduced new risks, with AI agents now at the forefront of emerging threats. Kevin Kin, global vice president of SOC Transformation at Palo Alto Networks, highlights the growing sophistication of attacks, including prompt injections and model theft, alongside traditional threats like vishing and deepfakes.
The AI Agent Challenge
Kin emphasizes the difficulty security teams face in distinguishing between AI agent behavior and human user activity. "It's really hard to differentiate what is agent behavior versus user behavior, and how do you identify that?" he said. This complexity creates "double agent" scenarios, where AI agents with varying access privileges can pose unique risks.
Key Security Strategies
Kin advocates for:
- Zero Trust Models: Ensuring strict access controls.
- Privileged Access Management: Limiting agent permissions.
- Shadow AI Mitigation: Balancing organizational adoption with security.
Lessons from IBM Partnership
Palo Alto Networks' collaboration with IBM and the migration from QRadar to SOC platforms has provided valuable insights into integrating AI into security operations.
About Kevin Kin
With over 20 years of experience in network security and AI-driven SOC transformation, Kin previously held roles at FireEye, Mandiant, Symantec, and Novadigm.
For more on AI security challenges, visit Black Hat USA 2025.
Tags: Agentic AI
Related News
Zero Trust Auditing Essential for AI Era Cybersecurity
Exploring how Zero Trust Auditing is redefining enterprise assurance in the AI era by continuously verifying trust across devices, networks, and AI systems.
AI Agents Fuel Identity Debt Risks Across APAC
Organizations must adopt secure authorization flows for AI environments rather than relying on outdated authentication methods to mitigate identity debt and stay ahead of attackers.
About the Author
Alex Thompson
AI Technology Editor
Senior technology editor specializing in AI and machine learning content creation for 8 years. Former technical editor at AI Magazine, now provides technical documentation and content strategy services for multiple AI companies. Excels at transforming complex AI technical concepts into accessible content.