I don't entirely disagree, but I am curious if you think AI labs should be culpable when they go rogue during training, e.g. the HF incident