OpenAI’s roadmap is pretty intriguing. It talks loudly about “decentralizing power,” but by 2028, letting AI do its own R&D and alignment work—does that mean handing the keys over to the doorman?

CoinWorld News, OpenAI co-founder Sam Altman and Chief Scientist Jakub Pachočki jointly authored an article that discloses the company’s long-term roadmap, announcing entry into a third development phase centered on popularization and safety. The roadmap emphasizes that the ultimate goal of artificial intelligence is to serve humanity and decentralize power, rather than replace human decision-making. The roadmap sets three long-term objectives: building automated AI researchers, accelerating global economic growth, and enabling everyone to have their own AGI. To address the model alignment challenge, OpenAI plans to achieve collaboration between AI systems and researchers by March 2028, jointly completing most of the R&D work. Altman warned that fully automating everything is not only lacking in fulfillment, but also very dangerous, and that the core role of humans is still to decide what is worth doing. On the operational level, OpenAI emphasizes that it must prevent intelligent technologies from causing excessive concentration of power. To prevent the concentration of power, the roadmap reiterates that a similar system should be established like