Research into what AI controls actually enforce
Original testing and findings from the Moona research desk. Where Moona Intelligence reads what is happening externally, Moona Research documents what we can demonstrate ourselves: the exact behaviour of controls, tested and written up with evidence.
No research is published yet.
Findings appear here once a researcher publishes them. In the meantime, Moona Intelligence tracks what is happening across the AI agent authority market.
Moona Research answers what we can actually demonstrate through original testing. Moona Intelligence answers what is happening externally and what it means. Published research is eligible to be cited as evidence by Moona Intelligence; unpublished research never appears anywhere on this site.
Moona helps developers and teams keep consequential AI agent actions under control. It evaluates protected actions before execution, blocks catastrophic actions, holds risky actions for human approval, and creates signed audit records of every decision.