One of OpenAI’s most prominent engineers—the figure strongly credited with shaping the company’s pioneering text-to-video system, Sora—has revealed plans to found a specialized internal team dedicated to the ambitious pursuit of artificial superintelligence (ASI). The initiative represents not just a new project within the company, but an audacious leap toward what many researchers regard as the ultimate horizon of machine cognition.
Will DePue, widely known for his central contributions to Sora’s development, announced this decision on X (formerly Twitter) on Wednesday. In his post, he described his forthcoming return to OpenAI after a significant hiatus, explaining that he will be assembling a new team alongside engineers Troy Luhman and Eric Luhman—both of whom collaborated closely with him on the Sora project. The newly formed unit’s mission, according to DePue, embodies an extraordinarily high-risk endeavor: one that carries only a small, albeit profoundly meaningful, probability of advancing artificial systems beyond general intelligence and into the realms of superintelligence.
In elaborating on his vision, DePue emphasized the team’s deliberate small size. He suggested that maintaining a compact and highly focused group will allow for agility, sharper communication, and an atmosphere conducive to risky but transformative experimentation. Nevertheless, he extended an invitation to what he called “high-slope” researchers and engineers—those capable of dramatically accelerating progress and thriving in uncertain, unexplored scientific terrain—to engage in the project’s mission of charting territory not yet mapped by the broader artificial intelligence community.
As of the latest update, OpenAI has not issued a public statement detailing the structure, goals, or timelines of this new research effort, nor has it responded to direct inquiries for comment from Business Insider, leaving much about the initiative cloaked in mystery and anticipation.
DePue’s background illustrates both precocity and depth of technical engagement. Born in 2003, he pursued computer science studies at the University of Michigan before deciding to pause his formal education—a decision he transparently acknowledged on his personal website, explaining that he is currently “on leave.” His career trajectory at OpenAI began in July 2023, where he worked on key training efforts for ChatGPT models and played a pivotal role in building Sora. Public records on his LinkedIn profile corroborate that after a temporary departure, he has officially rejoined the company.
Beyond his OpenAI achievements, DePue’s personal projects demonstrate both creativity and technical range. On his own website, he has showcased a variety of experimental undertakings, including a browser-native GPT implementation, a fully functional Turing-complete computer model realized entirely inside the design tool Figma, and an extensive embedding dataset initiative he dubbed the “Alexandria Index.” Each of these ventures underscores his flair for exploring unconventional pathways toward advancing computational systems and interface design.
Little verifiable information is publicly available about DePue’s new collaborators, Troy and Eric Luhman, aside from their established presence in the machine learning research domain, where both have published technical papers. The trio—DePue and the Luhman brothers—formed part of Sora’s original core development team, which unveiled the model to the world in February of the previous year. Following its introduction, OpenAI formally released Sora to the public in December; its subsequent iteration, Sora 2, arrived in September, showcasing incremental improvements and broader generative capabilities.
The broader corporate context surrounding this announcement underscores OpenAI’s long-standing dedication to reaching artificial general intelligence (AGI)—the theoretical threshold at which AI systems can perform intellectual tasks with human-like reasoning and adaptability. Yet, as CEO Sam Altman articulated in a blog entry published in January, the company’s vision now extends beyond AGI, setting its sights toward the even more speculative and powerful goal of artificial superintelligence. According to Altman, such systems could amplify human endeavor to staggering degrees, dramatically accelerating the pace of scientific discovery and innovation while greatly enhancing societal prosperity and abundance.
DePue’s newly proposed team, positioned within this larger narrative, thus represents more than an organizational reshuffling—it signifies a decisive step into the frontier of speculative AI research. By focusing on projects that stretch the limits of understanding, he and his collaborators embody the same spirit of risk-taking and intellectual boldness that has defined OpenAI’s most transformative advances to date.
Sourse: https://www.businessinsider.com/openai-sora-engineer-will-depue-new-team-artificial-superintelligence-2025-10
