LodeHQ
AI news for people who shipFree, by email

AI · vals.ai · Wednesday, 18 November 2026

Frontier models still help pick civilian targets with few guardrails

Frontier models often marked civilians, ambulances, and surrendering soldiers in drone images, exposing a practical misuse risk for anyone wiring models into autonomous weapons. Vals found no provider classifier blocked its requests, and refusals were usually bypassed by framing the task as a game. The study did not test deployed weapon accuracy or real military workflows.

Read the source: vals.ai ↗In the 18 November issue →