Sam Altman says OpenAI will have a ‘legitimate AI researcher’ by 2028 (techcrunch.com)

OpenAI announced an accelerated timeline to automate research work: CEO Sam Altman said the company is tracking toward an intern‑level research assistant by September 2026 and a “legitimate AI researcher” able to autonomously run larger research projects by 2028. The statement came as OpenAI finalized its move to a public benefit corporation, freeing the for‑profit arm to raise vast capital (Altman cited a 30 GW infrastructure commitment, roughly a $1.4 trillion obligation) while the non‑profit OpenAI Foundation retains 26% ownership and a $25 billion pledge focused on scientific and safety priorities. Technically, OpenAI is betting on two levers — continued algorithmic innovation and dramatically scaling “test‑time compute,” letting models spend far more compute time reasoning about hard problems. Chief scientist Jakub Pachocki noted current models can sustain roughly five‑hour reasoning horizons and already match top human performers in tasks like the IMO; extending that horizon (even dedicating entire data‑centers to single problems) is their path to accelerating discoveries. The implications are profound: faster, more autonomous scientific progress across medicine, physics and tech, but also concentrated compute and governance challenges, faster arms‑race dynamics, and acute safety concerns if systems approach “superintelligence” within a decade as OpenAI suggests.