If the Pentagon refuses to say, we need to look at the evidence ourselves:

1) Unprecedented scale and speed

2) A target database that appears not to have been updated since the school's separation from the IRGC complex in 2016

3) Confirmed use of AI to generate target packages (summaries explaining why a given target should be struck)

4) The Trump administration significantly loosening requirements for human supervision of the kill chain

5) Reproducible reasoning by Claude that relies on obvious visual markers (flat roof, fence etc.) but lacks the contextual judgement a human military analyst develops over years of experience and applies to subtle, ambiguous cases.