Can we trust OpenAI?
Its articles of incorporation define its mission as “advancing digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return." Sam Altman would later say, as to why OpenAI chose to be a nonprofit, "we don’t want to ever be making decisions that benefit shareholders. The only people we want to be accountable to is humanity as a whole."
OpenAI needs additional capital to keep building large AI models, so they decide to create a for-profit subsidiary. How is this compatible with their nonprofit history, and the donations they've received in light of it? The justification given is that the majority ownership, and therefore the primary oversight, remains in the hands of the non-profit entity, an organization legally required to prioritize the interests of society over private financial gain.
In late 2023, the non-profit board votes to fire OpenAI CEO Sam Altman for dishonesty. However, Altman and his allies, as well as Microsoft (a major investor in the for-profit) and employees whose equity relies on future fundraising for the for-profit, put up a fight. After a fierce contest between the nonprofit board and the investors and employees, Altman returns, and two of the board members who voted to fire him resign.
In June, rumors circulate that Sam Altman was considering changing OpenAI to a for-profit entity, and that he may be granted an equity stake in the company. By October, OpenAI officially began talks with regulators to convert to a for-profit entity.
The conversion would keep the non-profit entity in existence but, crucially, would give control over OpenAI’s operations and intellectual property to a for-profit entity. In exchange, the non-profit may be compensated with a non controlling equity stake (reported to be at least 25%) in the new for-profit company. The process to determine the value of the nonprofit's assets - namely control over the company and intellectual property - is being negotiated behind closed doors by OpenAI and Microsoft.
Whatever the nonprofit’s final compensation — even if it exceeds 50% — it will almost certainly be non voting shares. This means that the nonprofit will no longer be able to conduct oversight of OpenAI and carry out its original mission to “advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return.”
An Open Letter:
Originally published May 16, 2024