Yeah, when someone writes a vague prompt that simply says "Do what it takes to finish X" and the unleashed agents end up doing something lethal and illegal you would think its on the AI company to prevent any illegality and not the vague prompt writer.