AI news ·
OpenAI files for IPO and unveils three goals for its third phase
OpenAI filed for an IPO while unveiling a three-phase roadmap that targets automated AI researchers by March 2028 and a personal AGI for every person. The company also proposed a public wealth fund and a 32-hour workweek to distribute economic gains.
OpenAI entered its third development phase on the same day it secretly filed for an IPO, releasing a roadmap that sets three targets: building an automated AI researcher, accelerating economic growth, and delivering a personal AGI to every person on Earth. The company said AI systems could be collaborating with human researchers on full scientific projects by March 2028, while emphasizing that full automation is not the goal.
CEO Sam Altman and Chief Scientist Jakub Pachocki co-authored the post, titled "Built to Benefit Everyone: Our Plan," which reframes OpenAI's strategy as it moves toward public markets. The timing links a capital raise with a public vision statement - a pattern OpenAI repeated during its corporate restructuring last year.
The three-phase roadmap and what it means
OpenAI summarized its trajectory in three stages. Phase one was conducting research toward AGI. Phase two turned that research into products and observed how people used them. Phase three, where the company now operates, focuses on making AI accessible to all. The blog said the economy is being reshaped around AI, and the central question has become how to make advanced AI affordable, secure, and easy enough for everyone to use.
Three goals anchor this phase. First, create an "automated AI researcher" capable of independently driving large research projects. Second, accelerate economic growth by using AI to boost science and productivity while ensuring the benefits are widely shared. Third, give every person a personal AGI - a system individuals can direct toward their own needs, from understanding medical bills to starting a small business.
The thread connecting all three is straightforward: teach AI to do research, use that capability to accelerate society, and distribute the gains to every individual.
Automated AI researcher: the March 2028 target
Pachocki described the automated AI researcher as a system that can independently run large research projects - not just answer questions or write code. In a March interview with MIT Technology Review, he called this OpenAI's "north star" and said the company is close to building models that work continuously like humans, so that "you'll essentially be placing an entire research laboratory inside a data center."
The roadmap splits this into two milestones. By September 2026, OpenAI aims to create a "Research Intern" that handles several small projects within limited timeframes. By March 2028, the target is a "Full-time AI Researcher" that drives complete projects over the long term. Pachocki said the difference between the two is not intelligence but "how long a time span they can work independently."
This is not a superintelligence suddenly rewriting itself. It's closer to breaking the research process into stages and automating each one: AI writes code, tunes parameters, and filters options while humans step back to orchestrate the intelligent agents. The concept traces back to mathematician I.J. Good's 1966 prediction that a superintelligent machine could design better machines, triggering an intelligence explosion. AI improving AI has moved from theory into the lab - DeepMind's AlphaEvolve, a coding agent that automatically improves algorithms, is already used to optimize its own model training processes.
For professionals working with AI tools, this shift has implications. AI Research Courses cover the tools and methodologies that underpin this kind of automated scientific work, from model evaluation to iterative experimentation.
Economic growth and personal AGI
OpenAI's second goal - accelerating the economy - includes specific policy proposals. In April, the company released an industrial policy document that suggested creating a Public Wealth Fund jointly financed by government and AI companies. The fund would invest in AI growth and distribute returns directly to citizens as an "AI dividend." It also proposed piloting a 32-hour, four-day workweek, converting AI-driven efficiency gains into shorter hours without cutting wages.
The third goal, personal AGI, draws an analogy to rural electrification in 1920s America. Before electricity, daily life meant drawing water by hand, preserving food with ice, and ending the day at sunset. OpenAI argues AI should become as embedded in daily life as electricity became - accessible to everyone, used on their own terms.
OpenAI stated explicitly: "Fully automating everything is not the future we want, because such a future would be both empty and dangerous." The stronger AI becomes, the more important humans are in setting direction, making trade-offs, and bearing responsibility. "Deciding what is worth doing will, in the long term, remain humanity's most critical role," the post said.
Safety brakes and industry consensus
OpenAI called for an international coordination mechanism that could slow frontier AI development if safety, alignment, and societal resilience fail to keep pace. As early as 2023, the company suggested an organization similar to the International Atomic Energy Agency to oversee superintelligence.
Last week, Anthropic publicly stated that AI progress is moving too fast and leading labs may need to slow down or pause to let societal structures catch up. Two labs holding the world's most advanced models have now independently placed safety brakes on the agenda, moving the discussion from niche circles to an open industry priority.
The timeline for AGI itself remains uncertain. Musk bets on 2026. Anthropic CEO Dario Amodei said "powerful AI" could arrive as early as 2027. DeepMind's Demis Hassabis estimated around 2030, when "a system will emerge that deeply understands everything around you." Industry estimates cluster around the next five years.
Why this matters for IT and development professionals
OpenAI's roadmap shifts the question from "can we build AGI?" to "who owns it once it's built?" For developers, engineers, and infrastructure teams, the March 2028 target for AI systems that can independently run research projects signals a near-term change in how technical work gets divided between humans and machines. When AI can tune its own parameters, find its own errors, and iterate autonomously, the developer's role moves toward orchestration - setting objectives, defining constraints, and deciding what's worth building. AI for IT & Development training covers the automation patterns, coding agents, and infrastructure decisions that this shift demands. The more immediate question isn't when AGI arrives - it's whether teams are ready to work alongside AI that can improve itself.