Interesting research: “ AI Models Can Provably Hide Arbitrary Capabilities ” Abstract: Capability evaluations aim to surface dangerous model behaviors before deployment, but their reliability depends on the assumption that hidden capabilities can be elicited. We show this assumption does not hold under adversarial conditions by constructing backdoor attacks that embed encrypted neural networks…

برای اطلاعات بیشتر، متن کامل خبر را در منبع اصلی بخوانید: Schneier on Security ↗