> Reading between the lines, neither study has demonstrated that AI models are lying over their own volition, but instead doing so because they've either been trained or jailbroken to do so.
Correct, because to intentionally deceive one must be in possession of the truth (which LLMs arguably are incapable of).
Training sets for games of strategy should include behaviors considered deceptive and surprising. Is it any surprise then, that we see these behaviors reflected in the LLM's output?
