"OpenAI hacked Australian government website" - fixed the title.
I'm tired of people acting like these things are "agents" in the true sense of the word. If OpenAI unleashes a thing that has the potential to hack an entity, then OpenAI is culpable, not the "agent".