Speaking ahead of ADSTAR 2026, Defence’s chief data integration officer, Dr Paul Robards, and chief of space, intelligence, national security and cyber division, Andrew Seedhouse, outline how the Defence Force is engineering accountability into AI systems at scale.

The centrepiece is a philosophy that officers remain personally accountable for all decisions, with AI functioning as a high-powered analytical support tool rather than an autonomous agent.

The challenge is acute. Defence systems now ingest vast volumes of sensor data, satellites, drones, communications and cyber intelligence, creating information overload rather than clarity.

AI’s role is to parse this deluge, surface decision options ranked by strategic utility and present them with full traceability back to the source data. Commanders retain the final say; algorithms provide the why.

Both speakers emphasise the tension between technological velocity and governance rigour. Frontier AI labs release new capabilities weekly. Defence’s testing, security validation and policy alignment move faster than legacy defence timelines but must remain slower than pure research cycles. Experimental systems run on ranges with physical limiters; enterprise systems sit firewalled in sandboxes until rigorous vetting confirms fitness for operational deployment.

Equally critical is Australia’s industrial capacity challenge. Ukraine’s autonomous systems success reflects decades of manufacturing expertise. Australia must build equivalent capability rapidly yet cannot rely on Defence alone to sustain industry. Commercial diversification and allied procurement partnerships are essential.

The strategic upside for a middle power is stark: AI enables capability multiplication without conventional force parity. Small teams wielding superior analytics and autonomous systems can achieve disproportionate effect.

The hard work now is execution without ethical compromise.

Enjoy the podcast,
The Defence Connect team