Exactly. For all Anthropic and OpenAI's fearmongering about not releasing open weights AIs, because it will enable hackers - we have been seeing that hacking happening already. It is happening FROM THE CLOSED-WEIGHT LABS THEMSELVES