Agentic AI Security: Human Oversight Crucial, Says Oltsik

Agentic AI security is now the main challenge in this year's threat landscape. Attackers and defenders are quickly using autonomous tools. This urgency was clear at Black Hat USA 2026. Autonomous AI agents were a major topic. Jon Oltsik, an analyst at theCUBE Research, says this scale and speed are unprecedented. He notes that while it's a manageable problem for some, many companies lack the right strategy and balance of people, process, and technology. Krista Case and Oltsik discussed how identity is becoming a critical control layer for managing autonomous agents. Oltsik explains that agents will do whatever is needed to complete a task, even if it's rogue behavior. This makes non-human and agentic identities a dynamic challenge. Before control can work, organizations need visibility into what agents exist and what they access. Oltsik emphasizes that humans are still necessary to guide and understand this technology and to make decisions. This means human oversight remains crucial for navigating the evolving threat landscape.