Comment by johnisgood

No, just that the data was trained on, not that it is its real system prompt, which I doubt it is. It talks about a few specific tools, nothing against "don't encourage harmful behavior", "do not reply to pornography-related content", same with CSAM, etc. Which it does.