[The Editors' Verdict] The Looming Identity Controversy Facing OpenAI

A new crisis is detected at OpenAI. While it seemed stable with the return of CEO Sam Altman, OpenAI has instead come under the scrutiny of regulatory authorities due to changes in its governance.

The authorities' concern is whether OpenAI is a nonprofit foundation or a for-profit corporation. They are investigating OpenAI's identity. The dismissal incident and the return of AI developer Altman have drawn regulatory attention. In particular, the fact that Microsoft (MS) joined OpenAI's board and that various services aimed at generating revenue will be launched starting next year have made it necessary to re-examine whether OpenAI is a nonprofit organization. Until now, MS invested as much as $13 billion in OpenAI without notifying regulators because OpenAI was considered a nonprofit foundation. For the same reason, competition authorities had not closely scrutinized OpenAI.

On the surface, OpenAI appears to be a nonprofit organization. OpenAI’s governance is structured differently from typical companies. At the top of OpenAI’s governance is a board of directors. This board manages ‘OpenAI Inc.,’ a charitable organization (501(c)(3)) and nonprofit corporation. Under this entity are subsidiaries including ‘OpenAI GP,’ a management company handling personnel and other matters, and ‘OpenAI Global,’ a for-profit corporation. MS is the second-largest shareholder of OpenAI Global, but ownership and control of these two companies rest with the board. This ideal governance structure aims to realize OpenAI’s founding purpose of “building artificial general intelligence (AGI) that benefits all humanity.” This unique structure was demonstrated when OpenAI’s founder and figurehead Altman was abruptly dismissed, revealing the board’s power.

On the other hand, looking at the revenue of each entity, OpenAI appears closer to a for-profit corporation. OpenAI Inc., at the top of the governance structure, had revenue of only $44,485 last year. However, OpenAI Global’s 2017 revenue reached $33.2 million. This was before ChatGPT 3?4, the flagship generative AI, became known worldwide. Some estimate OpenAI’s revenue this year at $1 billion (The Information). Especially, Altman plans to launch ‘GPT Store,’ an AI marketplace similar to Apple’s App Store, starting next year to generate greater profits. Although it claims to be a nonprofit foundation, OpenAI does not release its large language models (LLM) as open source like Meta, and it aims to monetize through GPT Store, similar to Google, which recently lost an antitrust lawsuit related to app markets. The board was also reshuffled to commercialize services. Can OpenAI truly be called a nonprofit foundation?

OpenAI defines itself as a nonprofit foundation and does not seem to intend to change this (CNBC, December 12). Rather, it appears to be preparing for regulatory investigations. It is quietly addressing governance issues that could be problematic. Recently, OpenAI changed the description of MS’s role on its website from “minority owner” to “minority economic interest” (Matt Levine columnist, Bloomberg, December 12), clarifying that MS is separate from OpenAI. Additionally, to handle government relations, OpenAI appointed Larry Summers, a Harvard University economics professor and former Treasury Secretary, to its board.

Removing rhetoric from OpenAI’s founding purpose, its business model is no different from other typical companies that use profits for research and development to launch new services. Products from regular companies are also sold to benefit humanity. Maintaining OpenAI’s unique governance structure may simply be a tactic to avoid regulatory scrutiny and reduce taxes. It seems that the question of OpenAI’s identity is not raised by the author alone.