Nonprofits are generally considered a force for good, a beacon of hope in a capitalist system that places shareholder profits first and foremost. But what happens when nonprofits decide to go to the other side, in pursuit of the almighty dollar? A recent lawsuit filed by Elon Musk against OpenAI is shining a spotlight on this issue.
Established as a nonprofit back in 2015 by Musk and others, including Sam Altman, OpenAI’s mission, as stated in its charter, is to “ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity.”
Following conflicts over how to run the organization, Musk quit the board in 2018, and the following year OpenAI adopted a for-profit subsidiary. The lawsuit contends that, with the release of ChatGPT-4 in 2023, OpenAI abandoned its original mission of making its technology and research publicly available, claiming that it “has been transformed into a closed-source de facto subsidiary of the largest technology company in the world: Microsoft.”
While the suit’s grounds and Musk’s motives for filing it have been widely debated, examining the events that led up to it provides an interesting case study in what can happen when a nonprofit decides to adopt some sort of profit-making mechanism.
The stated motivation for OpenAI establishing a for-profit subsidiary was that it would enable the organization to raise additional capital to pursue its mission. Fundraising is a daunting challenge for any nonprofit, and judging by the results, OpenAI was light-years ahead of even the most sophisticated organizations. OpenAI’s nonprofit secured more than $130 million in donations, over $44 million of which Musk claims to have contributed. However large, this amount fell far short of the estimated $1 billion needed to hire the staff and build the massive computing power needed to achieve its long-term vision.
It is not unusual for nonprofits to seek revenue sources that supplement individual donations and grants. Many do so by creating a social enterprise that enables them to generate earned income or by charging a fee for their services. OpenAI decided to create a “capped-profit” company centered first and foremost on advancing the organization’s mission. Essentially, this new entity was a hybrid between a nonprofit and for-profit organization that enabled OpenAI to raise money from investors while limiting their returns. Control of the company was in the hands of OpenAI’s nonprofit board, which would ensure adherence to its charter.
The release of ChatGPT-4, along with Microsoft’s decision to increase its investment in OpenAI to $13 billion, put this hybrid business model to the test, and also forms the basis for Musk’s lawsuit. According to NPR, the structure of the new entity created four years prior had already caused frictions to brew between those in the company who remained committed to the nonprofit’s mission of serving the greater good and those with a Silicon Valley mindset of capturing market share. Some board members were said to be concerned that CEO Sam Altman was overlooking the risks associated with AI and rushing to bring OpenAI products to market.
On November 17, 2023, the nonprofit board fired Altman, citing a lack of transparency. Microsoft responded by extending an employment offer to Altman and any other OpenAI employee that decided to join its ranks.
A letter published four days after Altman's termination seems to confirm the frictions between the two camps within the company. Written anonymously by a group of former OpenAI employees, it goes so far as to claim that, in their belief, “a significant number of OpenAI employees were pushed out of the company to facilitate its transition to a for-profit model. This is evidenced by the fact that OpenAI's employee attrition rate between January 2018 and July 2020 was in the order of 50%.”
Are such conflicts the logical result of bringing profits into a nonprofit context? This is not always the case. There are many examples of nonprofits that have successfully adopted revenue generation models. The Girl Scouts sells cookies to support their activities and teach entrepreneurial skills. Unite to Light donates one solar light for every one sold. The Mayo Clinic generates revenue through a fee-for-service model while maintaining nonprofit status.
The original employees of OpenAI (the nonprofit) were likely largely driven by idealism—computer science PhD types more interested in advancing a groundbreaking tool to benefit humanity than growing their stock portfolios. As the technology advanced, its vast market potential became increasingly evident.
Was the adoption of a hybrid nonprofit/for-profit model a genuine attempt to preserve OpenAI’s founding philosophy? Or was it, intentionally or not, one step on a continuum towards becoming a fully for-profit company, akin to function creep?
Perhaps there are no guardrails strong enough to withstand a $13 billion investment by Microsoft (that’s $13,000,000,000). The sequence of events seems to suggest as much.
Following Altman’s ouster, over 90% of the company’s employees (702 out of 770) threatened to leave if Altman wasn’t rehired and a new board installed. He was and they were. (Current and newly incoming board members include leaders hailing from the Bill & Melinda Gates Foundation, Sony, Instacart, Salesforce, and Quora, and a former U.S. Secretary of the Treasury. The only board member predating Altman’s firing is Adam D’Angelo, the CEO of Quora. Sam Altman will also rejoin the board.)
In February of 2024, OpenAI announced that it would allow employees to sell their shares in the company, valued at $86 billion. That figure certainly poses a conflict of interest for even the most well-intentioned of organizations. To put it into context, even the top giving foundations award only a fraction of this amount every year, with the Gates Foundation topping the list at around $4 billion and the number three slot sliding drastically down to around $600 million.
Is it ethical, or even legal, for a nonprofit organization originally funded by individual donations to contribute to personal enrichment, to the tune of tens of billions?
Rather than directly addressing OpenAI’s nonprofit status, in legal terms Musk’s lawsuit is framed in terms of a breach of contract. Traditionally, only the state attorney general is able to sue a nonprofit for violating its mission.
However, there have been calls to do just that. Suggesting that OpenAI may now be under the control of its for-profit subsidiary, Public Citizen president Robert Weissman penned a letter to California Attorney General Rob Bonta asking him to “investigate whether OpenAI, Inc. is not acting to carry out its purpose and, if appropriate, to seek its dissolution, divestment of its assets, and reinvestment of those assets to charitable purpose.” It outlines the events following Altman’s firing to illustrate that, after this event, “there was effectively a fight between the non-profit entity and the for-profit entity and its stakeholders…and the for-profit won.”
The letter also stated that if OpenAI were to voluntarily convert into a for-profit or dissolve completely, as has been rumored, its assets should also be divested and allocated perpetually to a charitable entity, a requirement under the state’s laws. Other nonprofits turned for-profit in the state have had to abide by those requirements. For example, the California Endowment, which currently has assets exceeding $3 billion, was created when Blue Cross of California acquired a for-profit subsidiary.
Only time will tell if OpenAI is compelled to do the same, and whether or not it loses the lawsuit filed against it by Musk. The nonprofit community and OpenAI’s original donors, one in particular, will be watching very closely.