UC Berkeley’s Jesse Finfrock explains the pressures companies face to use AI to generate profit and the necessary considerations for entity formation.
The OpenAI leadership shakeup is generating welcome interest in corporate governance. The focus raises at least two key points to consider as business and society reshape around generative AI technologies.
First, companies face immense pressure to use AI to generate profit. This, weighed against the importance of ensuring health and safety, demands reevaluation of corporate governance priorities.
OpenAI’s founders initially recognized this by forming the entity as a 501(c)(3) public charity with a mission to “build general-purpose artificial intelligence that benefits humanity, unconstrained by a need to generate financial return.” Financial pressure apparently overwhelmed this concern, highlighting the role of corporate structure in balancing mission and profit.
Second, OpenAI’s dustup demonstrates that founders and legal counsel should take special care when determining entity formation. Of particular intrigue is OpenAI’s hybrid for-profit, nonprofit structure—or tandem entity, the term preferred by Tomer Inbar, a leading tax-exempt specialist at Morgan Lewis.
As more is learned about OpenAI’s decision-making, it will be valuable to keep in mind the following considerations that are unique to tandem entities.
Form Follows Function
Any corporate structure must enable the organization to achieve its goals. It is unfortunately common to create a beautifully complex structure that distracts from, or undermines, operational objectives.
Instead, a tandem entity must ensure that nonprofit and for-profit activities are accretive in value to each other. Susan Mac Cormac, a partner at Morrison & Foerster who focuses on this legal niche, distills this lesson into a memorable mantra: Form follows function.
A tandem entity can’t succeed if nonprofit and for-profit operations compete against each other. In addition, misalignment is certain if nonprofit personnel receive less compensation than those at its related for-profit, particularly when the for-profit is granting equity.
Risk also increases if for-profit use of a public charity’s intellectual property undermines the charity’s credibility. Or if, as OpenAI describes, certain funding assumptions prove flawed, the founding mission and vision may be jeopardized.
Private Benefit and Inurement
A core tenet of charitable organizations is that they don’t have owners like for-profit businesses do. Public charities hold assets in public trust that must be used to serve the public interest, furthering a charitable purpose. In return, they receive significant governmental benefits, the most obvious of which is tax exemption.
The IRS implements this tradeoff via restrictive rules prohibiting “private benefit” and “private inurement” in tax-exempt entities. These prohibitions embed the concept that a public charity can’t excessively benefit private parties.
In a tandem entity, a public charity can’t provide assets to the for-profit, or otherwise benefit investors in the for-profit, without receiving at least fair market value. This includes IP, which is notoriously difficult to value.
Presumably for these reasons, OpenAI stated in its 2021 Form 990 that it adheres to private benefit and private inurement prohibitions.
Specifically, OpenAI discloses including terms in its tandem entity agreements—the contracts that define the relationship between the nonprofit and for-profit entities—that “give priority to exempt purposes over maximizing profits for the other participants” and prevent it “from engaging in activities that would jeopardize the organization’s exemption.”
Compensation and Disqualified Persons
In connection with these prohibitions, there are strict limits on how public charities compensate personnel, especially those the IRS deems to be “disqualified persons”—people in positions of financial control at the organization, such as executive officers and directors.
As a general rule for public charities—private foundations face additional self-dealing restrictions—compensation must be “reasonable” and not more than “the value that would ordinarily be paid for like services by a like enterprise under like circumstances.” This includes non-cash benefits, such as equity compensation, issued by related for-profit businesses.
If OpenAI’s personnel were in control positions, they would likely be “disqualified persons” during the time they serve in such positions, as well as for a five-year lookback period in which these restrictions still apply. As has been reported, unease with the employee “phantom equity” structure—profit participation units—may have been a precipitating factor in the upheaval at OpenAI.
Conflicts of Interest
To help ensure adherence to the mesh of IRS rules, there must be no debilitating conflicts of interests among those forming, funding, or controlling the tandem entity. Best practice is for separate counsel—or firms—to represent the for-profit and nonprofit entities, including in early negotiations.
If equity compensation is involved, then it is best to engage outside valuation experts. Following formation, disinterested directors—those without conflicting financial or personal interests—should be appointed at both entities to represent their particular needs.
OpenAI asserts it complies with conflict-of-interest rules by requiring all contracts entered into with the non-profit to be on terms that are “at arm’s length or more favorable” to it.
Notably, OpenAI’s for-profit entity doesn’t have its own board but is managed by a separate entity wholly owned and controlled by the nonprofit. Presumably, this decision was made to protect tax-exemption; however, it adds pressure to the nonprofit board to responsibly manage both entities with care.
These and many other IRS rules governing tandem entities create a dense thicket of risk to be navigated by founders and legal counsel, lest they face accusations of misusing charitable assets or engaging in excess benefit transactions.
The early—perhaps regretful—decision to incorporate OpenAI as a public charity, followed by subsequent decisions to form and operate a profit-seeking business, reflect certain judgment calls. Tandem entities needn’t bear the blame for these decisions.
This article does not necessarily reflect the opinion of Bloomberg Industry Group, Inc., the publisher of Bloomberg Law and Bloomberg Tax, or its owners.
Author Information
Jesse Finfrock is an attorney and adviser teaching social enterprise law at University of California Berkeley School of Law.
Write for Us: Author Guidelines
To contact the editors responsible for this story:
Learn more about Bloomberg Law or Log In to keep reading:
Learn About Bloomberg Law
AI-powered legal analytics, workflow tools and premium legal & business news.
Already a subscriber?
Log in to keep reading or access research tools.