Oopsie-daisy from the king of AI himself: Sam Altman straight-up admitted OpenAI “screwed up” GPT-5.2 by laser-focusing so hard on making it a coding/engineering god that they accidentally turned its writing skills into sad beige mush. Flatter tone, worse translations, hallucinations in real docs, instant mode that forgot how to instant—it’s like they built a Ferrari brain but gave it the personality of expired oatmeal. The boss basically shrugged, “We prioritized smarts over soul, whoops, next version will fix it.” Classic tech move: chase benchmarks, lose the plot, then publicly confess like it’s casual Friday.
Sam Altman Says Oops, They Accidentally Made the New Version of ChatGPT Worse Than the Previous One
