What's this about?
US targeting records still classified the site as military despite years-visible school construction. Bloomberg says a rushed kill chain passed that stale classification through Maven and human approval.
Did the AI choose the target?
The report does not show an autonomous weapon choosing whom to strike. It describes personnel wrongly expecting an information-fusion system to flag stale data, amid time pressure and reduced civilian-harm review.
What's the community saying?
HN 473 points, 242 comments at selection time.
- Responsibility remains with commanders and institutions, not a software label.
- Automation bias can make a polished interface appear more authoritative than its inputs.
- The unreleased investigation limits independent verification of the causal chain.