IC Technology & Surveillance — 2026-05-09

NRO Director Flags AI Explainability as Major Concern for Satellite Intelligence

Speaking at the USGIF GEOINT Symposium in Denver on Wednesday, outgoing NRO Director Chris Scolese called AI "explainability" a "major concern" and "still an open area of research," per Breaking Defense. Scolese said NRO is expanding AI for satellite fleet management, including orchestrating a SpaceX-operated low-Earth orbit constellation he said will exceed human operators' management capacity as it grows. He described real-time sensor data fusion as the harder validation problem, noting that live conditions cannot be replicated by test the way satellite operations can. NRO is using an internal GPU cluster called the "Ultra-Dense Environment" for model testing, and Scolese called on industry to help develop validation methods.

Analysis
Scolese's public call for industry to develop validation methods NRO lacks, per Breaking Defense's attributed quotes from GEOINT, confirms the agency holds no internal methodology to translate into acquisition policy. NRO is unlikely to establish formal AI explainability requirements for acquisition programs by end of calendar year 2026, as concept validation, internal consensus, and regulatory drafting historically unfold over years. The explainability gap is sharpest for real-time sensor fusion, which unlike satellite operations cannot be replicated in test. The public framing may instead serve to pull commercial AI developers into NRO's vendor ecosystem and secure research budgets.

2 sources

AI explainability is a major concern for National Reconnaissance Office: Director - Breaking Defense
Chris Scolese on AI's Role in NRO's Space-Based Intel Capability Delivery - ExecutiveGov