Sources: after a deadly school strike in Iran, US Central Command changed its targeting process, integrated open-source data, and upgraded Palantir's Maven
The changes include refining processes for vetting targets and adding open-source data feeds to better track civilian movement.
Context & Ripple Effects
CENTCOM's AI use in the Iran campaign was initially framed around speed: reporting said Maven, integrated with Claude, helped identify and prioritize targets at scale, while CENTCOM described using AI to rapidly verify large volumes of operational data. Subsequent reporting instead pointed to disconnected military databases as a possible contributor to the February school strike and cited overreliance on Maven among contributing factors.
The reported changes recast Maven from a system associated with rapid target prioritization into one component of a more explicit target-vetting workflow, with open-source data intended to supply context on civilian movement.
First-order effects
- CENTCOM reportedly changes its lethal-targeting workflow by tightening target vetting and adding civilian-movement signals from open-source feeds.
- Palantir's Maven reportedly receives an upgrade as CENTCOM revises the process in which the system is used.
Second-order effects
- Maven's value to CENTCOM shifts toward supporting corroboration across intelligence sources, rather than relying on rapid AI-generated prioritization alone.
- The changes place greater operational importance on connecting data sources and on the human process that validates a target before action.
Third-order effects
- The episode points to a military-AI procurement standard centered on auditability, data integration, and safeguards around civilian harm—not merely the speed at which a system can produce targets.
- If this pattern holds, defense AI vendors will be judged as much on how their tools fit accountable decision workflows as on model performance or target throughput.
The trend: Military AI is moving from rapid target generation toward integrated, human-governed decision systems designed to make intelligence inputs more reviewable.