← Back to context

Comment by Iolaum

2 months ago

I 'd say he is an outstanding engineer as well. He may favor output over security more than outstanding engineers at 2025 but in the 2026 world what he does is impressive. And with OpenAI's resources he has turned OpenClaw's security woes around. Latest versions are much more secure than 2 months ago.

It’s much worse than that. Openclaw cosplays security, which is much more dangerous than just outright defining the security model as “this is new and risky, better sandbox this thing thoroughly at a different layer”.

It sucks at both security and usability as a result (all the vibe-designed security layers are constantly getting in my way).