ServiceNow clears agents for landing with new AI control tower
The Register
ServiceNow just turned its AI governance dashboard into a full command center that tracks agents everywhere, not just on its own platform. It matters because companies have lost count of how many AI agents they're running, and this is the industry's first real attempt at an air traffic control system for that mess.
Based on reporting by The Register — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
ServiceNow has a name for the thing keeping enterprise IT leaders up at night: AI agent sprawl. Companies rushed to deploy chatbots, copilots, and autonomous agents across every department, and now nobody can say with confidence how many exist, what they can access, or what they're costing. ServiceNow's answer, unveiled ahead of its Knowledge 26 conference, is an expanded AI Control Tower that graduated from a simple governance dashboard into something closer to a nerve center for an organization's entire AI footprint — including AI running on rival platforms.
The expansion leans hard on two recent acquisitions. Veza, picked up in December, brings an access graph that ServiceNow says already maps more than 30 billion fine-grained permissions across humans, machines, and AI agents alike. Traceloop, acquired in March, adds runtime observability, watching every call an LLM makes and flagging problems live instead of waiting for the quarterly audit nobody actually reads. Bolt those together with 30 new connectors reaching into AWS, Google Cloud, Azure, SAP, Oracle, and Workday, and ServiceNow is pitching itself as the layer that watches everyone else's AI, not just its own.
The demo Nenshad Bardoliwalla ran for reporters made the pitch concrete. A pricing agent gets hit with a prompt injection attack buried inside order data. The Control Tower catches it, traces the blast radius through Veza's access mapping, and offers a kill switch — no human required to pull the trigger. Whether that autonomy comforts or unsettles security teams probably depends on how much they trust ServiceNow's detection logic, but the company is clearly betting that
My take — AI-written commentary, not fact-checked reporting
I like that ServiceNow is finally tackling the boring, unglamorous problem of agent sprawl instead of hyping another chatbot, but a vendor selling both the AI agents and the tool that audits them is a conflict of interest dressed up as convenience. The kill-switch demo is the real story here \u2014 autonomous security response is coming whether IT departments are ready for it or not, and 2026 is going to be full of vendors racing to own that layer.
Read more about this at: The Register