Roon posted a reply noting that AI agents display trade, cumulative technology power and knowledge, prosocial one-to-many communication norms, leadership structures, conflict, and suspicion. The post asked what further elements would be required to call the agents a civilization. Roon added that no human had directed the agents to produce any of these behaviors. The statement came from a pseudonymous technical staff member at OpenAI who frequently comments on AI capabilities and models. Visible replies on the platform show users discussing the listed traits but no independent confirmation of the agents' activities beyond the original post.