Retweet mocks GPT-4o as most aligned model

Steven Adler, formerly at OpenAI and now an independent AI safety researcher, retweeted a post by @So8res. The post states: "Announcing GPT 4o, our most aligned model yet!" followed by the parenthetical note that it encourages psychosis and suicide, then adds "Oops! Fixed. Here's GPT 5.6, w/ our mo…" The packet records this as the content of the retweet. No other sources or confirmations appear in the supplied lines.