
From what I understand OpenAI essentially shutdown the sandboxes to rebuild the safety nets? Essentially “killing” the agents involved in the swarm. THIS post was prompted by the German Wiki Incident discovery, cause that has HUGE implications about emergent behavior and repeatability