The dream of a fully autonomous research engine has taken a sharp turn in the corridors of Silicon Valley. OpenAI, once vocal about its ambition to build a self-driving AI researcher by 2028, is now pivoting toward a “tandem” approach where humans and machines collaborate. CEO Sam Altman and chief researcher Jakub Pachocki have issued a stark warning: total automation is neither desirable nor safe. They are also calling for an international body with the power to throttle AI progress if necessary.
Last autumn, the company set a bold target. By March 2028, they intended to deploy an AI system capable of conducting research independently. That vision has been revised. A fresh blog post from Altman and Pachocki adopts a far more guarded stance.
“Our internal belief is that by March of 2028 we may have a significant fraction of our research being done by AI systems in tandem with our own researchers,” they write.
While the exact drivers behind this U-turn remain unclear, the message is clear: fully automating human tasks is a societal failure waiting to happen. Altman and Pachocki argue that as AI grows more powerful, the human role becomes more critical. We need people to set direction, make trade-offs, and apply judgment. As Altman put it, “Entirely automating everything is not the future we want. It would be unfulfilling, and it would be dangerous. […] A key long-term role for people will be deciding what is worth doing.”
OpenAI wants an international body that could slow AI development
Simultaneously, OpenAI is arguing that “AI doing AI research will become the determining factor of the pace of progress within the next few years.” Echoing the recent calls from Anthropic, the company is now floating the idea of slowing down development. Altman and Pachocki propose an international organisation to coordinate leading AI efforts and reduce catastrophic risks. It should enable “coordinated action, including slowing frontier development when needed, so societal resilience, safety, and alignment can keep pace.”
This request sits somewhat awkwardly alongside OpenAI’s own very ambitious goals. In the same blog post, the company outlines three core objectives: “Build an automated AI researcher,” “Accelerate the economy,” and “Give everyone on Earth a personal AGI.”
Altman and Pachocki describe the current moment as the start of OpenAI’s third phase. The first phase focused on foundational research. The second built the company into a product business. Now the goal is to make advanced AI “abundant, affordable, safe, useful, and easy enough” for every person and organisation to benefit.
“Frontier capability is only part of the job. The bigger task is turning that capability into tools people can actually use to thrive,” they write. That thinking lines up with OpenAI’s broader strategic shift from model provider to implementation partner, the same logic behind its new DeployCo subsidiary, which sends engineers directly into companies to embed AI into their workflows.
It is also a tacit admission that AI implementation is far harder than just offering a chatbot and that adoption may not move fast enough, with clear enough ROI, to support the revenue growth OpenAI and Anthropic will need in the coming months.
Key takeaways
- OpenAI has abandoned its goal of a fully autonomous AI researcher by 2028, replacing it with a plan for human-machine collaboration.
- Leaders are explicitly warning against total automation, arguing it would be dangerous and leave humanity without a say in what is worth doing.
- The company is calling for an international body that could pause frontier development to ensure safety and societal alignment.
- Despite these cautionary notes, OpenAI continues to pursue ambitious goals, including giving every person on Earth a personal AGI.
