NCSC Calls for Tight Security and Human Oversight as Agentic AI Use Expands

The UK National Cyber Security Centre warns organizations to adopt agentic (autonomous) AI cautiously, recommending incremental, low-risk pilots, strict privilege limits, strong monitoring, and clear human accountability to mitigate increased attack surface, unpredictable behavior, and oversight challenges introduced by autonomous decision-making and tool access.