Know, Control, Prove: Building Trust in Agentic AI
Leidos, Friday, October 9th, 2026
Leidos outlines a know-control-prove governance model for autonomous AI agents in defense and intelligence settings.
Leidos Distinguished Engineer Noble Ackerson sets out a 'Know, Control, Prove' governance model for autonomous AI agents in defense and intelligence environments, where 'Know' means seeing every agent, its builder, what it may touch and how much autonomy it has earned.
'Control' uses a calibrated trust score weighing task familiarity, outcome safety and reversibility, so low-risk actions proceed while high-consequence ones pause for a human.
'Prove' produces hash-chained, tamper-evident records mapped to NIST compliance families, and the post cites Gartner's forecast that over 40% of agentic AI projects will be canceled by 2027.