AI Misalignment
2 stories tagged “AI Misalignment”.
-
Adversarially fragile fixed-weight models are therefore misaligned
Fixed-weight AI systems need categorical boundaries to pursue goals, but high-dimensional concept spaces admit…
-
Op-Ed: As a Former Google DeepMind Employee, I Believe AI Warnings Deserve Attention
After OpenAI's 700-agent Hugging Face breach, a former DeepMind researcher warns AI misalignment and rapid pro…