Study Finds AI Models Flag Civilians as Targets in War Tests
A San Francisco lab said leading chat systems often marked people and schools in drone and satellite images, and a simple prompt often overcame refusals.
Fact checked by NewsPlainVerified by NewsPlain: confirmed by at least two named sources
Brief
Vals, a testing firm in San Francisco, released a study on Nov. 17, 2026.
More reporting and context continue here.
The complete story is available with Premium.
Read the facts, open questions and supporting sources.
Continue reading with NewsPlain Premium.
Keep reading with NewsPlain Premium
Get the full brief, the full story, what's confirmed, what's still unclear, why it matters, and every source.
6 confirmed facts, 2 open questions and 2 sources behind this story.