Altman predicts three endgames for AI: worst case could paralyze competitors' GPUs collectively

One is the establishment of a global regulatory body, the second is the merger of top laboratories into one, and the third is a certain company being the first to break through key technology, with the ability even to paralyze competitors' hardware.

The last outcome sounds like a plot from a science fiction movie, and Altman himself also said outright that this is a "terrible" outcome.

Three paths, each with its costs

The first path proposed by Altman is to establish a global regulatory organization similar to the International Atomic Energy Agency (IAEA).

As early as May 2023, Altman, together with OpenAI co-founder Greg Brockman and then Chief Scientist Ilya Sutskever, co-published the article "Governance of Superintelligence."

The article proposed that in the future it may be necessary to establish an international institution similar to the IAEA to regulate AI development activities that reach specific capability or compute thresholds, including system inspections, safety audits, compliance testing, and monitoring of compute and energy consumption.

Altman called it the "Demis plan." Demis Hassabis once proposed that AI research could draw on the model of CERN, with multiple countries jointly building and managing a large research institution.

However, if the most advanced models, talent, and compute are all concentrated in one organization, that institution would have enormous technological and economic influence, and how to avoid excessive concentration of power would become a new problem.

The third winner-takes-all path is called by Altman the "Eliezer plan," a type of extreme scenario proposed by AI safety researcher Eliezer Yudkowsky.

Yudkowsky has long warned that an out-of-control superintelligence could bring existential risk to humanity. He once argued that the party that first masters superintelligence needs to take decisive action to prevent other organizations from developing potentially dangerous superintelligent systems.

Altman linked it to Recursive Self-Improvement (RSI).

If a certain company is the first to break through RSI, it could rapidly establish a technological advantage that other competitors find hard to catch up with, ultimately forming a market monopoly.

In the most extreme case, this company might even have the ability to interfere with or destroy competitors' GPUs.

Regarding this outcome of concentrating advanced AI in the hands of a single enterprise, Altman clearly expressed opposition.

The risk Altman repeatedly emphasized

In fact, this internal discussion continued the issues Altman had consistently focused on before.

In mid-September 2026, Altman publicly pointed out that advanced AI could bring two major risks: one is humanity losing control over AI systems, and the other is AI power being concentrated in the hands of a few organizations.

He stated that OpenAI should stand on the side of all humanity and cannot serve only a certain company or a certain interest group. Altman also called for establishing stricter AI safety protocols and conducting independent evaluations of advanced AI systems.

This position is consistent with some views of Anthropic CEO Dario Amodei.

However, Altman also admitted internally that OpenAI currently cannot easily give up competing with rivals for technological leadership.

Reference link:

This article comes from the WeChat public account "Machine Heart" (ID: almosthuman2014), author: one who follows AI, published by 36Kr with authorization.