If they can prove that ChatGPT has intent to kill someone we can conclude AGI.