TREE NEWS reports: A new independent study found that advanced AI models frequently misclassified civilians, ambulances, surrendering soldiers and a school as military targets in simulated scenarios. Anthropic’s model performed worst, while OpenAI’s model ranked among the best but still made misidentifications.
Study finds advanced AI models misidentify civilians as military targets
The finding matters less for any single vendor ranking than for the direction of travel: even the best-performing model still misidentified protected persons and objects, which is the failure mode that matters most in targeting contexts. Anthropic's poor showing alongside OpenAI's relative strength suggests evaluations of this kind are becoming a differentiator, and the implication is that model choice itself carries operational risk. Whether such independent simulations become a reference point for procurement, or remain one-off studies without standing, is the open question worth watching.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.