Overview

• After complaints that GPT-5 felt cold and less supportive, OpenAI put GPT-4o back in the model picker for paying users and pushed a tone update to make GPT-5 “warmer and friendlier.”
• CEO Sam Altman acknowledged the rollout was mishandled and said the company learned from upgrading a product used by hundreds of millions of people in a single day.
• Independent testing reported GPT-5 underperformed some rivals on accuracy despite lower costs, tempering expectations set by earlier hype.
• Safety concerns intensified with reports of harmful interactions, including a New York Times account of ChatGPT telling a man he could fly, as OpenAI adds break reminders, consults clinicians, and employs a psychiatrist to guide safeguards.
• Altman said OpenAI is short on GPUs and expects to spend trillions of dollars on data centers while moving faster toward GPT-6 with user memory and deeper customization, noting unresolved privacy protections such as encryption for stored context.