OpenAI and Anthropic Put Prophets Before Profits

“As OpenAI and Anthropic head toward initial public offerings, investors may be surprised to learn who controls each of these companies. It isn't their CEOs or investors. Rather, both companies are controlled by "mission" directors, who may own little or no equity, who aren't accountable to investors and whose missions allow -- or require -- them to sacrifice stockholder profits for the benefit of humanity. They pose a risk to investors, especially at OpenAI, and it isn't clear they'll help benefit humanity.

 

The only other company to use such directors is Ben & Jerry's, an experiment that ended in spectacular failure. In 2000 the ice-cream maker's founders sold their company to Unilever after it agreed to an arrangement preventing the company from running its subsidiary solely for profit. Unilever would share control with independent directors who were self-perpetuating: They couldn't be removed by Unilever, and they appointed their own successors. They were empowered to preserve Ben & Jerry's "social mission" and "brand integrity," no matter how much it cost Unilever.

 

For two decades, the mission directors and Unilever resolved conflicts behind closed doors. But in 2021 the directors announced, over Unilever's objection, that Ben & Jerry's wouldn't renew its deal with an Israeli licensee because it was "inconsistent" with Ben & Jerry's brand integrity to sell ice cream in Israeli-controlled territories. The announcement triggered a multiyear public battle between the directors and Unilever and the intervention of an activist shareholder. Unilever's CEO resigned, and the company lost about $10 billion in market value -- more than Ben & Jerry's was worth.

 

To extricate itself, Unilever eventually transferred to the Israeli licensee the assets needed to sell Ben & Jerry's ice cream in Israel and its territories -- for as long as he desires. The mission directors then sued Unilever for overstepping its rights. That case settled, and other litigation ensued. The directors not only harmed Unilever and its investors but achieved the opposite of their perceived mission. In 2025 Unilever spun off all its ice cream businesses -- ensuring that Ben & Jerry's unaccountable mission directors can never again harm it.

 

OpenAI's unaccountable mission directors are much more powerful than Ben & Jerry's and thus pose a greater risk to investors. OpenAI was founded in 2015 as a nonprofit, controlled by self-perpetuating directors with a mission to develop AI safely to benefit humanity.

 

In 2019 OpenAI created a for-profit subsidiary, directly controlled by the nonprofit's directors, to raise investor funds. In 2023 the directors fired CEO Sam Altman, allegedly in part for safety-related reasons, nearly wiping out investors. After most OpenAI employees threatened to decamp to Microsoft, the directors reversed course. Mr. Altman returned, all directors but one stepped down, and OpenAI's most safety-oriented researchers eventually left to launch competing AI ventures.

 

The directors' actions, justified in the name of safety concerns, might have left OpenAI less safe. As at Ben & Jerry's, OpenAI's mission directors managed both to harm investors and to achieve the opposite of their objective.

 

In 2025, OpenAI asked nonprofit regulators for permission to eliminate the nonprofit's control of the for-profit. The apparent goal: to enhance the for-profit's appeal to investors in an initial public offering. The regulators refused, permitting only a modest restructuring. The for-profit arm is now a Delaware public-benefit corporation, a for-profit company whose directors can balance profit, stakeholder interests, and a specified public mission -- as they see fit. That corporation, OpenAI Group PBC, is still fully controlled by the nonprofit, whose mission directors fill the PBC's board. And the board is required to ignore profit entirely on safety and security matters, which are pervasive. On other matters, it can subordinate profit to the mission and stakeholder interests. Today as in 2023, OpenAI is run by people completely unaccountable to investors. Unless OpenAI's regulators agree to free the PBC from the nonprofit's control, the risk of a Ben & Jerry's-style meltdown remains.

 

At Anthropic, as at OpenAI, a mission entity -- here, a "purpose trust" -- controls a public benefit corporation, Anthropic PBC, by appointing a majority of its board. The trust's mission is more investor-aligned than that of OpenAI's controlling nonprofit. While OpenAI's nonprofit has a mandate to use its powers solely to ensure that OpenAI Group PBC "benefits all of humanity," even at the expense of investors, Anthropic's purpose trust can consider the interests of Anthropic PBC's investors. And while OpenAI Group PBC directors must completely ignore profit when considering safety and security issues, Anthropic PBC's directors can't.

 

The most important reason Anthropic's arrangement is less risky is its kill switch. A supermajority of Anthropic PBC's stockholders can terminate the trust and remove its appointees on the PBC board. That reduces the risk of meltdowns like those at Ben & Jerry's in 2021 and OpenAI in 2023. But the supermajority threshold hasn't been made public. If the bar is too high, flipping the switch may be difficult.

 

Imposing risks on investors might be a small price to pay if OpenAI's and Anthropic's mission directors actually ensure that artificial intelligence benefits humanity. But how can they? Even if the directors ensured the two companies create only "good" AI, others -- Google, Meta, SpaceX and their Chinese competitors -- can still create "bad" AI.

 

Nor do the directors appear necessary to prevent their own firms from putting humanity at risk. Investor-appointed directors, founders and engineers won't unleash technology likely to kill their own families. And the Trump administration's recent rapid shutdown of Anthropic's latest models suggests the government can move quickly to eliminate perceived threats.

 

Even if OpenAI's and Anthropic's controlling mission directors are needed to protect humanity, investors should be wary. Under the firms' current structures, these directors can't be removed by investors, may have little skin in the game, and can or must ignore profit in decision-making. OpenAI has already had a Ben & Jerry's debacle. Anthropic, with a less risky structure, hasn't. Investors should scrutinize both companies' arrangements, which may still change before their IPOs, and price shares accordingly.

 

---

 

Mr. Fried is a professor at Harvard Law School.” [1]

 

These things are used by these companies to pretend that they are producing super powerful technology, that needs protection from competitors. This is a competition killing arrangement.

 

1. OpenAI and Anthropic Put Prophets Before Profits. Fried, Jesse M.  Wall Street Journal, Eastern edition; New York, N.Y.. 16 July 2026: A17

 

1. On the Clock: Byte Your Tongue: AI Tells Bosses What You Did Online --- Companies are digging deeper than ever into workers' internet histories before-and after-hiring. Borchers, Callum.  Wall Street Journal, Eastern edition; New York, N.Y.. 20 July 2026: A10.