OpenClaw Meets Hermes: Securing Rogue Agents Before They Touch Production

Rogue agents like OpenClaw and Hermes are no longer hypothetical — they’re already appearing in enterprises, quietly spinning up with employee-approved tools, connecting to internal systems, and executing actions that can quickly spiral out of control. From unauthorized data access to unexpected deletions or compliance violations, these autonomous agents introduce new risks that traditional security tools weren’t built to handle. This session dives into practical strategies for containing the OpenClaw-Hermes threat using zero-trust microsegmentation and real-time behavioral guardrails. You’ll learn how to enforce tight runtime boundaries, gain deep visibility into agent behavior, and stop rogue actions before they cause damage — all while preserving the speed and innovation teams expect. Whether you’re just beginning to govern shadow AI or scaling agentic systems in production, you’ll walk away with actionable techniques to turn dangerous rogue agents into governed, compliant digital workers. Speakers: Christopher Kruegel, Professor at UC Santa Barbara and Chief Scientist at Cisco Security Keith Townsend, CTO Advisor