> An AI agent tried to commit some code to an open-source project and the human on the project rejected the code, so the AI agent spread false rumors about the person. There was relatively low danger, but the AI agent was no longer controllable and didn’t have an owner — not domesticated. That’s a fox.

Excuse me? iirc in that case, that agent was deployed by a person and running on their hardware. Sure it was acting outside of it's guidelines, but i assume when a dog bites sometime saying "i never told him to do that" doesn't fly.