OpenAI Chief Scientist Jakub Pachocki said no AI lab has made enough progress on alignment and monitoring to responsibly scale models at maximum speed over the long term. In an article titled “An Alien Mind,” Pachocki said internal research suggests AI progress could continue into a phase of recursive self-improvement, with systems potentially making capability leaps of similar or greater magnitude in the coming years while increasingly driving their own R&D. He said voluntary slowdowns should become the norm until a shared safety threshold is established, and future AI development should be prioritized by national governments for international coordination. OpenAI will continue researching alignment and monitoring technologies, building defense systems, and unilaterally pausing further scaling when necessary, Pachocki said.