OpenAI calls for an international body that can slow frontier AI development

Why it matters

OpenAI is pairing a March 2028 target for AI-assisted research with a call for global slowdown powers, exposing the conflict between frontier competition and collective safety.

In an X post on Tuesday, OpenAI co-founder and CEO Sam Altman and chief scientist Jakub Pachocki renewed OpenAI's call for an international organization capable of coordinating a slowdown in frontier AI development if advances begin moving faster than safety work and society can absorb.

The post resurfaced a plan published by Altman and Pachocki on June 8th, 2026. Its central premise is that AI systems will soon play a larger role in AI research itself, compressing the time needed to test ideas, find errors and develop more capable models. OpenAI says that feedback loop could eventually become fast enough that "the world will need to pace" frontier development.

Pachocki has worked at OpenAI since 2017 and previously led the development of GPT-4 and OpenAI Five, according to OpenAI's account of his appointment as chief scientist. He holds a doctorate in theoretical computer science from Carnegie Mellon University. His co-authorship puts OpenAI's top research executive alongside Altman in setting a timetable for the next phase of the organization's technical work.

OpenAI's March 2028 target

OpenAI's most concrete forecast is an internal target for March 2028. By then, Altman and Pachocki wrote, AI systems may perform a "significant fraction" of OpenAI's research while working alongside human researchers.

OpenAI did not define that fraction as a percentage. The plan describes an automated AI researcher that can increasingly handle parts of the research process while remaining steerable, accountable and connected to people. The distinction matters because software that improves the research process can accelerate every subsequent model cycle, rather than increasing capability in a single product release.

The automated researcher is one of three goals Altman and Pachocki set for what they call OpenAI's third phase. The others are accelerating scientific and economic progress and giving everyone a "personal AGI." OpenAI describes the third phase as a shift from building frontier models and consumer products toward making advanced AI broadly available, affordable and useful.

That distribution strategy is already tied to OpenAI's capital requirements. On March 31st, OpenAI said it closed $122 billion in committed capital at an $852 billion post-money valuation. Amazon, Nvidia and SoftBank anchored the round, with Microsoft participating. SoftBank co-led alongside Andreessen Horowitz, D. E. Shaw Ventures, MGX, TPG and accounts advised by T. Rowe Price.

OpenAI explicitly described that financing as fuel for more compute, faster research and wider deployment. OpenAI also said ChatGPT had more than 900 million weekly active users, its APIs processed more than 15 billion tokens per minute and revenue had reached $2 billion per month. Those figures are self-reported and were not accompanied by audited financial statements in the announcement.

The result is a direct institutional tension. OpenAI is amassing capital and computing capacity to accelerate model development while asking governments and competing frontier labs to prepare for a point when coordinated restraint may be necessary. The proposal reflects the pressure facing every frontier developer: slowing alone risks surrendering technical and commercial ground, while unrestricted competition could leave safety work behind.

A global coordination problem

Altman and Pachocki argue that an international organization should coordinate leading AI developers, establish shared safety expectations and enable collective action when catastrophic risks rise. Any institution with that mandate would need participation from competing companies and governments, access to credible information about frontier training programs, and a way to determine when a slowdown is warranted.

The commercial and national incentives run in the opposite direction. OpenAI is competing with well-capitalized model developers for researchers, customers, computing infrastructure and product distribution. Governments increasingly treat frontier AI as strategic infrastructure. A coordination body would therefore have to produce enough trust for participants to disclose risks without creating an advantage for rivals that remain outside the arrangement.

OpenAI frames broad access as part of its safety strategy, arguing that concentrated control over powerful AI would make society more fragile. OpenAI's own scale, however, makes the governance question immediate. The organization seeking wider distribution of AI capability is also building one of the largest concentrations of capital, compute and consumer reach in the technology industry.

The March 2028 research target gives that debate a deadline. OpenAI is telling policymakers and competitors that AI-driven research acceleration is an operational objective inside its current planning horizon, rather than a distant scenario. Its proposed international body would need to exist, earn cooperation and develop intervention rules before the systems it is meant to govern begin materially accelerating their own development.