\n<\/p>\n
OpenAI CEO Sam Altman said Tuesday the company is \u201crolling back\u201d the latest update to the default AI model powering ChatGPT, GPT-4o, after complaints about strange behavior, in particular\u00a0extreme\u00a0sycophancy.<\/p>\n
\u201c[W]e started rolling back the latest update to GPT-4o last night,\u201d Altman wrote in a post on X<\/a>. \u201c[I]t\u2019s now 100% rolled back for free [ChatGPT] users and we\u2019ll update again when it\u2019s finished for paid users, hopefully later today. [W]e\u2019re working on additional fixes to model personality and will share more in the coming days.\u201d<\/p>\n Over the weekend, users on social media blamed the updated model, which arrived toward the end of last week, for making ChatGPT overly validating and agreeable. It quickly became a meme. Users posted screenshots of ChatGPT applauding all sorts of problematic, dangerous<\/a> decisions<\/a> and ideas<\/a>.<\/p>\n On Sunday, Altman acknowledged<\/a> the problem and said that OpenAI would work on fixes \u201cASAP\u201d and would \u201cshare [its] learnings\u201d at some point.<\/p>\n<\/div>\n