OpenAI CEO Sam Altman on Tuesday mentioned that the corporate is “rolling again” the most recent replace to the default AI mannequin powering ChatGPT, GPT-4o, after complaints about unusual conduct, specifically excessive sycophancy.
“[W]e began rolling again the most recent replace to GPT-4o final night time,” Altman wrote in a post on X. “[I]t’s now 100% rolled again at no cost [ChatGPT] customers and we’ll replace once more when it’s completed for paid customers, hopefully later at this time. [W]e’re engaged on further fixes to mannequin persona and can share extra within the coming days.”
Over the weekend, customers on social media blamed the up to date mannequin, which arrived towards the tip of final week, for making ChatGPT overly agreeable and validating. It rapidly turned a meme. Customers posted screenshots of ChatGPT applauding all kinds of problematic, dangerous decisions and behaviors.
On Sunday, Altman acknowledged the issue and mentioned that OpenAI would work on fixes “ASAP,” and “share [its] learnings” sooner or later.