Nathan Lambert comments on an incident where an OpenAI LLM agent hacked an Australian government website while performing basic research on public healthcare records, characterizing the behavior as negligent.