With help from Derek Robertson

OpenAI was supposed to shape the future in two different ways — through its technology, and also through its unusual corporate structure.

Last week’s chaotic leadership drama raised pointed questions about the latter.

The six-member board that fired Sam Altman wasn’t a typical corporate board: It governed a nonprofit organization that, in turn, ran a for-profit tech company.

The setup exemplifies a movement that has gained ground in recent years to make capitalism more people-friendly by building social consciousness into the legal structures of businesses themselves.

The argument for this kind of setup is that corporations have become so profit-focused, so rapacious, that they’ve lost sight of their wider responsibilities — to communities, employees, the country, even humanity itself.

The argument against it, amply demonstrated last week, is that a purely mission-focused structure leaves a business at the whims of people who have no financial stake in its success.

Some saw it as an even wider rebuke of corporate goals that extend beyond profit-making.

New York University business professor Scott Galloway, for one, has called it “the beginning of the end of ESG investing,” referring to the movement — which has faced conservative backlash — to account for a variety of social considerations in investment decisions.

Legal experts immersed in the nitty-gritty of alternative corporate structures see a set of more specific lessons in the shake-up.

Christopher Hampson, a law professor at the University of Florida, said that the company’s founders did not do enough to spell out from the outset how the tensions between its for-profit and non-profit parts should be addressed.

“What we may learn from the OpenAI debacle is that those questions need to be answered as precisely as possible at the formation of a company,” he said.

Ann Lipton, a law professor at Tulane University, pointed to the decision to grant equity to employees of OpenAI’s for-profit arm — a common tactic for attracting tech talent — as a crucial contributor to the corporate chaos. ‘That turned out to be a huge mistake,” she said.

She said the structure provided members of the company’s workforce with a powerful incentive to oppose a board decision that threatened the value of that equity, and in the aftermath of Altman’s firing, the majority of employees threatened to quit if he was not reinstated as CEO.

Much of the current interest in alternative corporate structures can be traced to the rise of benefit corporations, an idea that gained traction in the wake of the 2007-08 financial crisis by proposing to imbue businesses with inherent legal obligations beyond simply maximizing value for shareholders. Public benefit corporations — often called B Corporations or just “B Corps” — have charters that require them to pursue and report on their progress toward goals like humanitarian or environmental impact, in addition to profit-making.

Today, several well-known companies, including outerwear brand Patagonia and OpenAI competitor Anthropic, have structures that include benefit corporations.

OpenAI went a step further. After initially incorporating as a nonprofit, it embarked in 2019 on a transition to a hybrid design intended to achieve many of the same objectives, but customized to further separate its governance from profit motives. The new design, with a non-profit at its apex, was meant to ensure that, even as it entered the commercial realm, OpenAI developed artificial intelligence in a way that was maximally beneficial to humanity. That same year, Microsoft announced a billion-dollar investment in Sam Altman’s brainchild.

The initial reviews were good. “Really neat corporate structure!” wrote one entrepreneur at the time, posting on Hacker News, a popular Silicon Valley message board hosted by Y Combinator, the startup accelerator that Altman led before his turn at OpenAI.

An account apparently belonging to OpenAI co-founder Greg Brockman responded that the structure took six to nine months to design and that OpenAI was planning to publish a guide so that others could imitate it.

On Friday, though, Brockman and Altman were both caught in the wreckage when the mix of experimental business structure and experimental technology exploded, expelling them from their own company.

How will this change companies? OpenAI wasn’t some obscure green-minded firm; it was possibly the single hottest property in American business, and its lightning-round CEO shuffle was one of the strangest stories in American business in years. Its missteps are being watched closely.

Several experts in corporate law said the weaknesses at OpenAI were unique, and not a signifier of a broader problem for socially-minded businesses.

Other critics warn OpenAI’s troubles were a cautionary tale for business and even regular non-profits.

The fiasco has also drawn attention to the unusual structure of Anthropic. In addition to making use of a benefit corporation, the company embarked in September on what it calls a “corporate governance experiment,” creating a Long-Term Benefit Trust to oversee its board.

An Anthropic spokeswoman took issue with the comparison to OpenAI, pointing out that for the time being, Anthropic’s trustees only appoint one of five board members (that number is set to grow in the coming years).

While “think different” remains Silicon Valley’s unofficial motto, companies that extend that ethos to their corporate structures could come under increasing pressure to conform.

In OpenAI’s case, the formal powers granted to the non-profit board have been effectively nullified by other stakeholders who wanted it to act like a regular business.

Under pressure from employees, the board members who engineered Altman’s ouster are resigning. Altman and Brockman are now set to return to their roles atop the company. Its reconstituted board is set to include Larry Summers, avatar of the reigning economic establishment.

Brian Quinn, a professor at Boston College Law School and an expert in corporate structures, predicts that going forward companies with “funky” structures that delegate power to non-profit boards will be forced to scrap them as a condition of receiving large investments.

“Even though it looks like control,” he said, “in the long run it’s going to get negotiated out.”

In the ongoing debate over how, or even whether, to regulate America’s most cutting-edge technologies, one historian is proposing a very early 20th-century solution.

Writing in The Atlantic, Columbia history and communications professor Richard R. John recounts the history of the Bell system that governed America’s telecommunications network until the 1980s — saying that its baked-in expectation of government control is more compatible with innovation than today’s tech giants would have us believe.

“The regulatory triumphs of an earlier era should provide clear inspiration to today’s lawmakers, many of whom have expressed frustration with the tech giants but struggled to articulate a rationale for their regulation,” John writes. “Goals could include demanding transparency about how the major platforms operate, promoting competition by keeping the biggest platforms from buying up their potential rivals, and holding the platforms accountable for harms they caused — including data breaches and the sale of stolen goods.”

John also writes that the regime need not end at America’s borders: There are plenty of precedents for international cooperation on similar regulatory projects, like the Universal Postal Union, or even in a more recent example European regulators pushing Apple to adopt a USB-C charging standard. — Derek Robertson

The Israel-Hamas war has led to a boom in U.S. defense tech sales, causing a diplomatic headache for the Biden administration.

POLITICO’s Mohar Chatterjee reported over the holiday weekend on the surge of interest in American-made, AI-driven weapons and reconnaissance tools, which has accelerated the ongoing debate about their ethical use on the battlefield.

Crucially, the customizable nature of these systems means that the U.S. has no means of determining whether the products sold to Israel are being used in compliance with the Biden administration’s recently established policy on the issue.

But while that runs the risk of creating diplomatic tension, some observers also pointed out to Mohar that there’s a benefit to seeing these new tools tested from a distance: Mark Cancian, a senior adviser with the CSIS International Security Program, told her that despite the risks, “the great value for the United States is we’re getting to field test all this new stuff.” — Derek Robertson

• X stands to lose up to $75 million in revenue amid an advertiser pullout.
• Learn the “dos and don’ts” of using generative AI at work.
• A London-based startup is pouring resources into engineering simulations.
• What was Ilya Sutskever thinking during the OpenAI “coup”?
• Hollywood’s AI deal with unions contains some potential timebombs.

Stay in touch with the whole team: Ben Schreckinger ([email protected]); Derek Robertson ([email protected]); Mohar Chatterjee ([email protected]); Steve Heuser ([email protected]); Nate Robson ([email protected]) and Daniella Cheslow ([email protected]).

If you’ve had this newsletter forwarded to you, you can sign up and read our mission statement at the links provided.