Sam Altman Outlines OpenAI's Third Phase Plan

Per a blog post by Sam Altman and Jakub Pachocki, OpenAI says it is entering a "third phase" that prioritizes making AI abundant, affordable, safe, useful, and easy for wide adoption, Business Insider reports. The post describes the first phase as research toward artificial general intelligence and the second phase as releasing products and learning from how people use them. The authors list three stated goals: "to build an automated AI researcher, to accelerate the economy, and to give everyone on Earth a personal AGI," per the blog post quoted by Business Insider. The post also states, "Powerful systems must remain safe, aligned with human intent, and subject to human control," and warns that "entirely automating everything is not the future we want." Altman and Pachocki also said OpenAI believes 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," per Business Insider. Several outlets reported the post appeared as OpenAI filed confidential IPO paperwork with the U.S. Securities and Exchange Commission, framing the third-phase plan alongside a move toward public markets (Marketing-Interactive; Business Insider).

Companies and research organizations that move from capability-first research to broad deployment typically shift engineering effort toward robustness, scaling, and developer ergonomics. For practitioners, that often means greater emphasis on reproducible evaluation, toolchains for safe rollout, and production-ready APIs rather than only frontier-model benchmarks. Industry tooling trends over the past three years show rising investment in observability, red-teaming, and fine-grained access controls as usage expands at scale.

Industry reporting frames OpenAI's stated emphasis on an "automated AI researcher" and "personal AGI" as long-horizon ambitions rather than immediate product roadmaps; Business Insider attributes these phrases to the blog post by Altman and Pachocki. Observers tracking similar strategic statements note the recurrent pairing of broad capability goals with repeated public commitments to safety and coordination, and with calls for international governance structures in high-risk domains.

Indicators an observer can follow include concrete technical releases or papers that materialize the "automated researcher" concept, changes to developer APIs or pricing that advance "abundance," and any published governance proposals or participation in multilateral coordination fora. Business Insider's report reproduces the blog post's stated aims; the post offers a tentative internal timeline (a significant fraction of research done by AI systems by March 2028) but few concrete operational milestones for the broader goals.

The summary uses verbatim quotes drawn from the blog post as reported by Business Insider, including "Now we are entering the third phase" and the safety language noted above.