" If you ask ChatGPT or Claude to write something like this through their websites, they will refuse. This OpenClaw agent had no such compunctions."

It's likely that the author was using a different model instead of OpenClaw. Sure OpenClaw's design is terrible and it encourages no control and security (do not confuse this with handwaving security and auditability with disclaimers and vibecoded features).

But bottom line, the Foundation Models like OpenAI and Claude Code are the big responsible businesses that answer to the courts. Let's not forget that China is (trade?) dumping their cheap imitations, and OpenClawdBotMolt is designed to integrate with most models possible.

I think OpenClaw and Chinese products are very similar in that they try to achieve a result regardless of how it is achieved. China companies copy without necessarily understanding what they are copying, they may make a shoe that says Nike without knowing what Nike is, except that it sells. It doesn't surprise me if ethics are somehow not part of the testing of chinese models so they end up being unethical models.