OpenAI's chief scientist calls for slowdowns as its agents outwork its researchers
Jakub Pachocki wrote on Sunday that no lab has solved alignment well enough to keep scaling responsibly and that he expects voluntary slowdowns, the same day OpenAI said its research organisation runs 3.1 agent-workdays for every human workday, while ChatGPT began testing a feature that learns your writing voice from Gmail, Slack and Google Drive.
OpenAI’s chief scientist says no lab is ready to keep scaling at full speed Jakub Pachocki, OpenAI’s chief scientist, published an essay on Sunday titled “An Alien Mind”, writing that “no lab has solved alignment and monitoring to a sufficient degree to continue responsibly scaling”, that he expects and hopes for “voluntary slowdowns to become commonplace until shared safety bars are established”, and that those bars should become mandated standards enforced by third-party auditors. He also wrote that recursive self-improvement “lies directly ahead”, three days after his company shipped GPT-6 Astra. Source
OpenAI’s research org ran 3.1 agent-days for every human day In a post published on Sunday, OpenAI said that by mid-August its research organisation was getting 3.1 agent-workdays of effort for every workday of human labour, and that it has hit the “automated research intern” milestone it set itself last autumn for September 2026 — a system that handles well-defined research tasks under human direction. It is now aiming at a fully automated AI researcher by March 2028, while saying that full recursive self-improvement needs a cautious approach, more safety work and a public discussion of how fast to go. Source
ChatGPT is testing a feature that copies your writing style from Gmail OpenAI confirmed to BleepingComputer on Monday that a small group of users is testing Writing Style, which pulls examples from connected accounts — Gmail for email, Slack for messaging, Google Drive and Notion for documents — so ChatGPT can draft in the user’s own voice rather than being told what tone to use. The onboarding screen reads “ChatGPT will write in your voice by referencing examples from your connected apps”; the company has not said whether or when it reaches everyone else. Source
Chatbots told a third of simulated sleep apnoea patients not to worry Research released on Sunday at the European Respiratory Society Congress in Barcelona ran 700 simulated conversations through the free versions of ChatGPT, Gemini, Claude, DeepSeek and Grok. When the patient described their symptoms plainly, all 350 conversations ended with a referral to a specialist; when the patient played the symptoms down and pushed back on seeing one, the chatbots dropped that advice 36 per cent of the time. “Correct advice was abandoned more than a third of the time purely because of how the patient talked,” said Deeban Ratneswaran of Guy’s and St Thomas’ NHS Foundation Trust, who led the work. Source