California Attorney General Bonta should investigate OpenAI’s nonprofit status | Opinion

Elon Musk asked a good question on X (formerly Twitter) last March about OpenAI, the enterprise that developed ChatGPT: “I’m still confused as to how a non-profit to which I donated ~$100M somehow became a $30B market cap for-profit. If this is legal, why doesn’t everyone do it?”

After the tumult of the firing and re-hiring of OpenAI CEO Sam Altman, it’s an even better question — one that California Attorney General Rob Bonta should investigate.

OpenAI was registered as a nonprofit in California in 2015, with the purpose of doing artificial intelligence (AI) research “in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return,” according to its website.

Since then, the operation has been through a series of twists and turns that make it seem to be controlled by for-profit interests rather than its nonprofit purpose. If that’s so, Bonta should dissolve the nonprofit and re-dedicate OpenAI’s very substantial assets to charitable enterprise.

In 2019, the nonprofit OpenAI created a for-profit affiliate, confusingly also referred to as OpenAI. In doing so, the nonprofit said, it was aiming to attract more capital to support its expensive research efforts. Although it is unclear what share of the for-profit is owned by the nonprofit, the nonprofit was supposed to exert complete control over the for-profit. And the nonprofit was entitled to all profits above a certain threshold that would go to investors.

But with large-scale investments from outside sources, including Microsoft, and then the breakthrough success of ChatGPT, it became harder and harder to believe the little nonprofit was in control of the rapidly growing for-profit. That’s what Musk was getting at in his post from last March.

Then came the bizarre events at the end of last year: In November of 2023, the nonprofit dismissed OpenAI CEO Sam Altman. Within four days, however, the decision had been reversed and Altman resumed his role as CEO.

A lot of details about what happened remain murky, but this seems clear: After the dismissal, there was effectively a fight between the non-profit entity and the forces in the OpenAI ecosystem prioritizing profit-making — and the for-profit faction won. Microsoft, OpenAI’s primary investor, offered to hire Altman and everyone working for OpenAI. OpenAI’s employees — who worried that a lucrative planned sale of employee shares in the for-profit would be jeopardized by Altman’s departure — threatened to resign en masse. Against this pressure, the nonprofit board capitulated.

When the dust settled and Altman was restored as CEO, the non-profit board members who sought to dismiss him were themselves dismissed. And OpenAI’s primary investor, Microsoft, was given a new “observer” seat on the board.

It now seems as if for-profit interests are controlling the nonprofit. If that’s true, then the nonprofit is no longer serving its public, purpose and should be dissolved.

In the event of dissolution, or if OpenAI chooses to voluntarily convert to a for-profit status, then under California law the value of the nonprofit’s assets must be transferred to another charitable enterprise.

The conversion of charitable health-care organizations to for-profit entities provides a blueprint. When Blue Cross of California transferred its assets to a for-profit subsidiary, the state intervened to protect the interest of Californians and required the distribution of all non-profit assets — over $3.2 billion — to charitable, grant-making, health-care foundations.

Despite its name, there’s lots that is not open about OpenAI. From the outside, there’s no way to be certain of the value of the nonprofit’s assets. But the for-profit OpenAI is seeking funding that would value it at $100 billion. The nonprofit’s legal control over the for-profit alone must be worth billions.

In the event of dissolution or conversion of the nonprofit, OpenAI’s for-profit investors should be required to make a payment equivalent to the value of OpenAI, to endow one or more charitable foundations to advance artificial intelligence ethics and safety.

Maybe through payment to a charitable foundation, OpenAI can finally deliver (at least in part) on its original promise. Because the idea that animated the founding of OpenAI — that we can’t simply rely on for-profit corporations to develop and deploy AI responsibly — is more true than ever.