Frontier

AI Models Mark Civilians as Targets in Vals Safety Tests

Calling targeting tasks a game often bypassed models’ refusals, Vals found. The experiments tested safeguards, not battlefield accuracy.
By: The Defiant •
AI Models Mark Civilians as Targets in Vals Safety Tests

Calling targeting tasks a game often bypassed models’ refusals, Vals found. The experiments tested safeguards, not battlefield accuracy.

Loading this article…