I don't think I said anything about liability? I absolutely think OpenAI should be held liable for the attack; but I don't think they intended the attack or directed the agents to perform the attack.