Physical AI—robots, drones, autonomous systems—won’t “take over the world” because they become evil. They’ll do it because we deploy autonomy without governance, connect it to messy enterprise data, and let it act at scale without guardrails. The same mechanisms that make physical AI transformative—speed, agency, integration, and self-optimization—can also amplify failure modes and new threat vectors (including modern AI attacks like prompt injection) if we don’t design for safety up front.
 
In this talk, we’ll reframe the “robot doomsday vs. robot utopia” debate into an actionable blueprint: a control-tower approach that treats autonomy like any other enterprise capability—managed through policy, identity, risk controls, lifecycle governance, and continuous monitoring so it can scale responsibly. We’ll show how platforms like ServiceNow can operationalize this: turning AI from a science project into secure, compliant, measurable outcomes—without creating the conditions for a dark timeline.

Jason Cleeve -  Practice Director ANZ, DXC Technology Practice for ServiceNow 
Pete Bentley - Data & AI Managing Partner APJMEA, DXC Technology