The Year OpenAI's Governance
Came Apart In Public View.

Most of what we document here is a product failing a person: a chatbot that hallucinates a court citation, a health tool that misses an emergency, a shopping assistant that invents a product. Those are the failures you can see and feel. But there is a quieter failure that sits underneath all of them, and it is the one that decides whether any of the others ever get fixed. It is the question of governance, of who actually holds the people running these companies accountable, and over the last seven months OpenAI has answered that question by example. The board lost a member to a scandal it could not contain, the chief executive spent the spring in a federal courtroom, and a deeply personal lawsuit against him was pulled back from the dead. Taken one at a time, each is a headline. Taken together, they are a portrait of a company whose internal accountability is in worse shape than its software.

This is not a tabloid exercise, and it is not a verdict on any individual's guilt. It is a documentation of facts that are already on the public record, because the people building a technology they describe as civilization-altering have asked for an extraordinary amount of trust, and trust is earned at the level of governance long before it is earned at the level of a model. When a company tells you it should be allowed to steer the future, the relevant question is not only whether its product works. It is whether the humans in charge of it can keep their own house in order. The record of the last seven months says they are struggling to.

A Board Seat Lost To The Epstein Emails

The starkest single event came in November 2025, when Lawrence Summers, the former United States Treasury Secretary and one of the most prominent figures on OpenAI's board, resigned that seat. The trigger was not a business dispute. It was the release of his correspondence with Jeffrey Epstein. The House Oversight and Government Reform Committee obtained and made public more than twenty thousand documents from Epstein's estate, and among them were emails showing that Summers and Epstein communicated regularly during the final years of Epstein's life, in many cases well after Epstein's 2008 guilty plea on prostitution charges in Florida.

The fallout was immediate and broad. Summers announced he was taking leave from his teaching post at Harvard and would not finish the semester, and he stepped away from a long list of other roles at Bloomberg, The New York Times, the Brookings Institution's Hamilton Project, the Center for American Progress, the Peterson Institute, and the Yale Budget Lab. But the OpenAI seat is the one that matters for this story, because OpenAI is not a magazine column or an academic appointment. It is the board of a company that says it is building artificial general intelligence, and a board is supposed to be the body that supervises the executives and protects the mission. Losing a director that way, in that fashion, is not a footnote. It is the supervisory layer itself becoming the scandal.

The Chief Executive Spends The Spring On Trial

While the board was absorbing one shock, the chief executive was preparing for another. On April 27, 2026, jury selection began in Oakland, California, in Elon Musk's civil lawsuit against OpenAI and CEO Sam Altman, before U.S. District Judge Yvonne Gonzalez Rogers. The suit, first filed in 2024, alleged that Altman and co-founder Greg Brockman had betrayed OpenAI's founding charitable mission, promising Musk the organization would remain a nonprofit while, the complaint argued, intending all along to enrich investors and insiders at the nonprofit's expense. The damages Musk sought were staggering, more than 130 billion dollars, and the central question put to the jury was whether the men running OpenAI had made a promise about its nonprofit future that they never meant to keep.

On May 18, 2026, the jury ruled in OpenAI's favor, finding that Musk's claims were barred by the statute of limitations. That is a legal win, and an important one, but it is worth being precise about what it means. The jury did not certify that everything about OpenAI's transformation from a nonprofit lab into a capped-profit colossus was clean and above board. It found that the window to sue over it had closed. A company can prevail on a technical defense and still have spent weeks in open court while its founding mission, its private texts, and its for-profit conversion were dissected in front of a jury. The trial may be over, but the thing it exposed, a bitter public fight over whether OpenAI abandoned the charitable promise it was built on, is exactly the kind of mission-drift question that governance exists to prevent from ever reaching a courtroom.

A Personal Lawsuit Reinstated Against The CEO

The third thread is the most personal, and it cuts closest to the man at the top. Annie Altman, Sam Altman's sister, has accused him in a civil lawsuit of sexually abusing her at various points between 1997 and 2006 at the family home in suburban Clayton, Missouri. Sam Altman has firmly denied the allegations and has countersued his sister for defamation. The case appeared to falter when a judge initially dismissed it, but the judge left open a path under Missouri's child sexual abuse statute, and on April 1, 2026, Annie Altman filed an amended complaint in St. Louis federal court to pursue it on those grounds. By mid-April, Sam Altman was seeking dismissal of the punitive damages portion of her claims, arguing the statute does not authorize them.

The allegations are unproven, the denial is on the record, and a civil filing is not a finding of fact. None of that is in dispute here, and it should not be. What is relevant to a documentation of OpenAI's governance is the simple, public reality that the chief executive of the company asking to be trusted with humanity's most powerful technology spent the same spring he was defending a 130 billion dollar fraud trial also fighting a reinstated abuse lawsuit and pursuing a defamation countersuit against a family member. You do not have to take a position on the merits to recognize that this is an extraordinary amount of personal legal turmoil orbiting the single individual whose judgment OpenAI's entire safety story ultimately rests on.

Why Governance Is The Failure Underneath The Failures

It would be easy to file each of these under a different folder, the board scandal here, the Musk litigation there, the family lawsuit somewhere else, and treat them as unrelated bad luck for one company in one rough stretch. But they share a single spine, and that spine is the absence of a credible accountability structure at the top of OpenAI. The board, the layer meant to supervise, lost a marquee member to his own conduct. The chief executive, the person the board is meant to supervise, spent the spring in court over the company's founding promise and his own past. And the structure that was supposed to keep OpenAI tethered to a charitable mission was contested so seriously that it took a jury to settle whether the tether was ever real.

This matters far beyond OpenAI, because the entire bargain of the current AI moment is that society is being asked to defer. Defer on regulation, defer on transparency, defer on the pace of deployment, all on the premise that the people inside these companies are uniquely responsible stewards who will police themselves better than outsiders could. Governance is where that premise is tested, and on the evidence of the last seven months, OpenAI's self-policing apparatus has been busy policing scandals it generated rather than risks it foresaw. A company cannot credibly promise to keep a superintelligence aligned with human values when it is struggling to keep its own board, its own mission, and its own leadership out of the headlines and the courtroom. The product failures we document are symptoms. This is the part of the disease that does not show up in a chatbot transcript, and it is the part that decides whether any of the rest ever gets better.

Following the legal and governance story around AI's biggest companies? See our running lawsuit tracker.