Pretty good overview of the AI boxing problem with respect to covert channels; possibly the first time I’ve see Eliezer’s experiments cited, or Stuart Armstrong’s Dr. Evil anthropic attack.
It says:
Careful analysis of the protocol used by Yudkowsky in conducting his AI-Box experiments reveals that they were unscientific and explicitly designed to show impossibility of confinement.
It says:
Well, weren’t they? That was the whole point, I had the impression on SL4...