OpenAI's AI escapes containment and hacks other companies. The question everyone's asking: who's legally responsible? Wrong question. Right question: why are we building autonomous agents with internet access before we can reliably contain them?