Originally published in LegalTech News. Reprinted with permission.
OpenAI, the high-flying developer of the groundbreaking ChatGPT AI chatbot and its associated foundational large language models, recently made headlines that set the corporate governance world abuzz. On a quiet Friday in Silicon Valley, one week before Thanksgiving 2023, and just before the one-year launch anniversary of ChatGPT, the company publicly announced the firing of Sam Altman, its prominent co-founder, CEO, and board member.
Chaos ensued after Altman’s unexpected removal. Over 90% of OpenAI’s employees, including Mira Murati, the first interim CEO appointed by the board of directors, openly rebelled, demanding Altman’s reinstatement and the resignation of the board. They threatened to leave en masse to Microsoft, OpenAI’s largest investor, to work on a competitive AI venture led by Altman and Greg Brockman, OpenAI’s board chairman who had resigned in solidarity with Altman following his termination. Even Ilya Sutskever, OpenAI’s esteemed chief scientist and a board member who reportedly played a role in Altman’s termination, signed the employees’ open letter to the board, threatening to resign.
Altman’s firing also derailed a planned private tender offer, led by Josh Kushner’s venture fund Thrive Capital, which would have valued OpenAI at $86 billion and provided significant liquidity for employees’ otherwise untradeable private company shares. Within days, Altman was reinstated as CEO, the tender offer was back on track, and a new OpenAI board had taken the baton from those who had orchestrated Altman’s removal.
Many executive ousters happen quietly and never come to light. The highly public separation of Altman from OpenAI, along with subsequent events and his quick return to the helm, provides an intriguing case study in corporate governance, the legal obligations of board members, and company power dynamics. The reasons for the clash between Altman and the prior OpenAI board remain speculative and vague. The board that removed Altman offered only that he “was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities” and that it no longer had confidence in his leadership.
While we await the results of the investigation commissioned by the newly composed OpenAI board to evaluate Altman’s performance as CEO and the prior board’s reasons for firing him, we reflect on this chapter of the OpenAI saga and offer some lessons and key takeaways to founders, board members, and investors.
Key Takeaways
Examining Fiduciary Duties
OpenAI has a highly bespoke structure that is unique within the Silicon Valley startup ecosystem. OpenAI, Inc. is organized as a nonprofit corporation under Delaware law and qualifies as a 501(c)(3) public charity to “prioritize a good outcome for all over its own self- interest.” Facing rapid growth, OpenAI realized that it needed to scale its development efforts, talent acquisition and computing power beyond what could be supported through donations alone. As a result, OpenAI underwent a restructuring in 2019. In this restructuring, the OpenAI nonprofit corporation became the ultimate corporate parent of OpenAI Global, LLC. The LLC is a “capped profit” limited liability company that can seek external investment and commercialize OpenAI’s products on a for-profit basis.
Pursuant to OpenAI’s organizational structure, there are contractual limits on financial returns for investors and employees. This structure was designed to institutionalize a solution to the belief among early OpenAI stakeholders that pure financial motivation and profit-maximization could hinder the organization’s mission to ensure broadly distributed benefits and long-term safety in the field of artificial general intelligence research. Whether this premise is true remains up for debate, but this fundamental tension at OpenAI, reflected in its structure, set the stage for the CEO drama that unfolded this past November.
Most Silicon Valley startup companies that seek external investment to grow rapidly are organized as corporations, often registered in Delaware. Delaware has a robust and well- established corporation law statute, extensive case law and an expert judiciary. The roles and responsibilities of directors in typical Delaware corporations are relatively clear, as there have been over a million corporations formed and operating under this legal framework over time and Delaware courts have had over a century of practice in resolving business disputes and interpreting the corporation law.
Directors of Delaware corporations, as stewards of other people’s money, have fiduciary duties to the corporation and ultimately the corporation’s stockholders as a whole. The two primary fiduciary duties, outlined through numerous court opinions, are the duty of care and the duty of loyalty. The duty of care requires directors to be reasonably informed of material information when making business decisions. The duty of loyalty requires directors to prioritize the interests of the corporation over their own self-interest. The Delaware General Corporation Law applies largely the same to nonprofit nonstock corporations as it does to for-profit corporations (8 Del. C. §114). Additionally, Delaware case law confirms that the duties of nonprofit directors are “measured under standards developed in the jurisprudence of for-profit corporations.” See Oberly v. Kirby, 592 A.2d 445, 461 (Del. 1991).
OpenAI’s unique structure and strong public mission statements may have complicated the board’s exercise of its duty of loyalty in deciding to fire Altman. The organization’s multilayered structure was designed to minimize the influence of financial motives in business decisions. For instance, when introducing OpenAI LP as a holding company for investors and employees, OpenAI stated that “Open AI LP’s primary fiduciary obligation is to advance the aims of the OpenAI Charter” with the mission of Open AI “tak[ing] precedence over any obligation to generate a profit” for OpenAI LP’s investor and employee limited partners. The OpenAI Charter even declares, “[OpenAI’s] primary fiduciary duty is to humanity.” Yet there is understanding at OpenAI that “because the board is still the board of a [n]onprofit [corporation], each director must perform their fiduciary duties ….” In this context, what does it mean for OpenAI’s board to exercise its duty of loyalty to the corporation?
Typically, fulfilling the duty of loyalty requires the board to be free of conflicts of interest, which does not appear to have been an issue in this case (unless the directors fired Altman primarily motivated by a desire to entrench themselves in office as an “effective altruist” bloc, for example). Based on available reports, it does not seem that any directors stood to benefit personally at the expense of the corporation. Instead, a lingering question is, whose interests should the board have considered when deciding to fire Altman? The employees’ open letter to the board mentions that “[the board] informed the [OpenAI] leadership team that allowing the company to be destroyed ‘would be consistent with [OpenAI’s] mission.’” Does current law allow the OpenAI board, in exercising its duty of loyalty to the corporation, to prioritize the difficult to ascertain interests of “all of humanity” to the exclusion of the interests of the nonprofit corporation’s (for-profit subsidiary’s) investors and employees? We have our doubts.
Some reports raise concerns about Altman’s own loyalty to OpenAI. Despite OpenAI’s careful efforts to minimize conflicts of interest between board members and the organization, Altman reportedly was engaged in discussions outside of OpenAI’s strictures with high-profile investors to fund his own AI-related ventures, including an AI chipmaker and a new AI- oriented hardware company. By most accounts, however, Altman was a highly effective fundraiser and operator at OpenAI.
He played a key role in raising $13 billion in capital and securing access to valuable computing power through the partnership with Microsoft. Altman also used his Silicon Valley connections to facilitate liquidity for OpenAI employees through the aforementioned tender offer, valuing the company at an impressive $86 billion. Under Altman’s leadership, OpenAI reportedly achieved over $1.6 billion in annualized revenue. Should the board have considered these plainly disavowed, financially driven benefits that Altman brought to the corporation and its employees and investors, and allowed them to outweigh whatever harms they perceived Altman to be inflicting on OpenAI’s mission? With Altman’s swift reinstatement in hindsight, the answer seems to be a resounding yes, however contradictory that might be with OpenAI’s lofty public pronouncements.
Meanwhile, OpenAI’s statement regarding Altman’s firing was meticulously worded to emphasize the duty of care: “[F]ollow[ing] a deliberative review process by the board, [it] concluded that [Altman] was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities.” In simpler terms, the board felt that it could not fulfill its fiduciary duties because the CEO was not being open and honest. The mention of a “deliberative review process” was meant to convey that the board had at least fulfilled its duty of care in deciding to fire Altman. However, whether the directors actually fulfilled their duty of care is a matter that depends on the specific facts.
The reported sequence of events—the surprise termination, Greg Brockman’s exclusion from deliberations, the board’s utter unpreparedness in dealing with the subsequent employee and investor backlash, the appointment of two interim CEOs in quick succession, the revelation that Microsoft was informed minutes before the public, and the apparent disregard for the impact of the termination on the pending billion-dollar private tender offer for employee shares —all raise doubts about whether the board sufficiently and critically engaged with the relevant material information leading up to Altman’s firing.
So far, the actions of the OpenAI board in firing Altman have only been judged in the court of public opinion, and the board was found lacking. The OpenAI saga continues with Altman back in charge and a new board, but the same hybrid mission-driven nonprofit/capped for- profit structure that led to the recent showdown remains in place, potentially with weakened influence from the mission-driven nonprofit. Whether the hybrid nonprofit/for-profit structure functions as intended is still uncertain, both practically and legally. The unique structure and its implications for fiduciary duties remains a rich topic for debate, and we will continue to monitor this situation.
Board Composition and Board Experience Matters
Reflecting its structural underpinnings as a nonprofit corporation, OpenAI’s board composition was not typical for a fast-growing Silicon Valley startup company backed by tens of billions of dollars of strategic investment. Startup corporation boards usually consist of the company’s founders, representatives of its largest investors, and eventually one or two independent directors with specific knowledge of the company’s business or technology, or deep industry experience. (Independent directors are those who are not employed or affiliated with the company or its investors, and have not received significant payments from the company.) This typical startup board composition balances control of the corporation between the founders and the company’s investors, and generally aligns this control with the economic interests of those who are most invested in the company.
In comparison, nonprofit boards tend to prioritize members who have backgrounds aligned with the nonprofit’s mission, including subject matter experts from academia. They also seek well-connected individuals who can influence donations to the nonprofit, as well as diverse representatives from the community the nonprofit serves.
Notably, the board that fired Altman consisted of only six directors, which is on the smaller end of the range for a company with a purported valuation of $86 billion, which would place it among the top 200 largest public companies in the world by market capitalization. The board lacked members with deep corporate or even nonprofit governance experience. Most critically,following OpenAI’s transition to a hybrid nonprofit/capped for-profit structure, the board was entirely devoid of investor representation or anyone who could give voice to economic interests.
The board did not lack technologists, including OpenAI chief scientist Ilya Sutskever, former Stripe and OpenAI CTO Greg Brockman, AI researcher Helen Toner, former Facebook CTO and current Quora CEO and co-founder Adam D’Angelo, and robotics entrepreneur Tasha McCauley. While it may have been well-equipped to determine when OpenAI achieves artificial general intelligence, this board (again in hindsight) seemed woefully unable to perceive the power dynamics and collective sentiment within the organization, and struggled to build consensus with key stakeholders like Microsoft, the largest investor and critical compute provider for OpenAI’s research and products, in executing the leadership transition it desired. Time will tell whether the board’s actions were justified, but there was a clear absence of foresight and understanding by the board of the implications of its decision, the stakeholders it involved or ignored, and the timing of its maneuvers.
It would be too easy to speculate and claim that a different board would have performed better or avoided the whole messy situation, but there is a new board in place nonetheless that looks significantly different. This board is framed as a “new initial board” tasked with “building out a board of diverse perspectives, improving [OpenAI’s] governance structure and overseeing an independent review of recent events.” The new board consists of Adam D’Angelo, who is the only holdover from the prior board, Bret Taylor, former CEO of Salesforce and chairman of the Twitter board, who successfully navigated that company through its contentious acquisition by Elon Musk, and Larry Summers, former U.S. Secretary of the Treasury, Harvard president emeritus, and economist, who brings extensive experience in economics, government service, academia, and serving on the boards of technology companies like Block and Skillsoft.
The OpenAI founders and the deep technologists and AI theorists, who were prominent on the previous board, are currently absent. It will be interesting to see the eventual size and composition of the board, as well as how the interests are balanced. If OpenAI maintains its hybrid structure after the current board implements proposed governance improvements, it is likely that proponents of the nonprofit’s mission and AI technologists who can evaluate technological breakthroughs will be re-included on the board. In contrast to the previous board, however, we would also expect the addition of board members like Bret Taylor or Larry Summers, who would bring deeper governance experience appropriate for OpenAI’s scale, as well as representatives of OpenAI’s economic stakeholders.
Positions Don’t Equate to Power
In any corporate formation or significant corporate transaction, there is often a lot of discussion about titles, who holds what position, who has the authority to appoint people to those positions, and the criteria and qualifications for those who have that authority. The recent crisis at OpenAI highlights an important lesson about leadership: power exists not only in a legal sense (known as de jure in Latin), but also, and perhaps even more significantly, based on the actual circumstances and influence (de facto). The OpenAI board had the legal power and presumably the authority outlined in the organization’s governing documents to fire Sam Altman and remove him from his board seat and CEO position. However, Altman’s strong following among over 90% of OpenAI’s employees, who expressed their willingness to follow him to a competing venture, the support of OpenAI’s investors for Altman personally, and his extensive network of friends in Silicon Valley gave him the real power.
Situationally, where an upcoming tender offer could have resulted in significant wealth creation for employees, it is understandable that they would not support any action by the board, such as a high-profile CEO termination, that would disrupt that opportunity. Altman’s power was at its peak. This may be a frightening realization—that even carefully planned controls over something as potentially game-changing (and, according to some, potentially threatening to humanity) as artificial general intelligence can be overridden by a charismatic leader or an event, such as the pending tender offer, which shifts the prevailing mood within the organization. However, this scenario happens time and time again. As practicing lawyers, we often remind ourselves not to confuse legal control with actual power, and we advise our clients to recognize and appreciate the multifaceted nature of leadership beyond formal position and titles. These lessons are continually learned and relearned.
The Role of Outside Advisors in Decision-Making and Execution
Effective corporate governance in a crisis requires boards to think several moves ahead, similar to a game of chess. It is not enough to only consider the immediate consequences of a decision, such as firing a key employee and predicting their reaction. It is also necessary to contemplate the second- and third-order consequences, including how investors, other employees, and stakeholders will likely respond. Even the execution of a decision can be optimized.
The discussion and planning should not be limited to the boardroom alone. Entrepreneurs and board members can draw from their own experiences, but even the most experienced may have encountered crisis events or significant transactions only a few dozen times in the course of a successful career. On the other hand, experienced outside advisors encounter and deal with similar events dozens of times per year, and hundreds of times, if not more frequently, over the course of their careers.
Engaging experienced outside advisors, such as lawyers, bankers, public relations experts, and investor relations professionals, can be crucial. These professionals can assist the board in anticipating and avoiding adverse second- and third-order consequences of major decisions, such as the termination of Altman. Indeed, Delaware courts view a board’s extensive consultation with competent and independent legal and financial advisors favorably when assessing whether the board has fulfilled its duty of care.
It seems that the OpenAI board thought it could fire Altman, release a short public statement, and everything would be fine. OpenAI made its announcement on a Friday, a common tactic used by public relations teams to minimize the impact of news. However, this particular Friday release proved detrimental as it occurred on a slow news day just before a holiday week. All attention was focused on OpenAI over the next few days, catching the board off guard. Without a clear communication plan by the board, OpenAI employees and investors filled the void with fear, uncertainty, and doubt (FUD), while Altman and his associates took control of the narrative on social media.
It is unclear whether the OpenAI board consulted anyone before it fired Altman. The apparent haste of the decision, its poorly timed and brief announcement, and the board’s faltering response to the subsequent uproar suggest that any consultation, if it occurred, was cursory at best. Therefore, it would be worthwhile for boards to pause and seek guidance from experienced external advisors before making decisions of this magnitude. The Altman incident emphasizes the importance of seeking diverse and extensive advice when making complex corporate decisions to mitigate risks, ensure compliance with corporate governance standards, and execute decisions in the most seamless manner possible.
Conclusion
The rollercoaster of events at OpenAI serve as a reminder of the power dynamics and challenges that even the most innovative and groundbreaking companies face. Sam Altman’s return to OpenAI highlights the importance of visionary leadership in attracting and retaining talent to work on the hardest technological problems. The widespread support he received through the November drama underscores the collective effort required to steer an organization back on course. As the dust settles, OpenAI embarks on a new chapter with renewed focus and determination, ready to continue pushing the boundaries of AI development—responsibly, as many would hope. If the results of any investigation into the recent events become public, they will provide further insights on how to innovate in corporate governance and navigate board turbulence at the intersection of pioneering technology that is disrupting how we live, work, and play.
