Ex-OpenAI staffers file amicus brief opposing the company’s for-profit transition

A gaggle of ex-OpenAI workers on Friday filed a proposed amicus temporary in assist of Elon Musk in his lawsuit in opposition to OpenAI, opposing OpenAI’s deliberate conversion from a non-profit to a for-profit company.

The temporary names twelve former OpenAI workers: Steven Adler, Rosemary Campbell, Neil Chowdhury, Jacob Hilton, Daniel Kokotajlo, Gretchen Krueger, Todor Markov, Richard Ngo, Girish Sastry, William Saunders, Carrol Wainwright, and Jeffrey Wu. It makes the case that, if OpenAI’s non-profit ceded management of the group’s enterprise operations, it could “basically violate its mission.”

A number of of the ex-staffers have spoken out in opposition to OpenAI’s practices publicly earlier than. Krueger has known as on the corporate to enhance its accountability and transparency, whereas Kokotajlo and Saunders beforehand warned that OpenAI is in a “reckless” race for AI dominance. Wainwright has stated that OpenAI “shouldn’t [be trusted] when it guarantees to do the suitable factor later.”

In an announcement, an OpenAI spokesperson stated that OpenAI’s nonprofit “isn’t going wherever” and that the group’s mission “will stay the identical.”

“Our board has been very clear,” the spokesperson advised Trendster through electronic mail. “We’re turning our current for-profit arm right into a public profit company (PBC) — the identical construction as different AI labs like Anthropic — the place a few of these former workers now work — and [Musk’s AI startup] xAI.”

OpenAI was based as a non-profit in 2015, nevertheless it transformed to a “capped-profit” in 2019, and is now making an attempt to restructure as soon as extra right into a PBC. When it transitioned to a capped-profit, OpenAI retained its nonprofit wing, which presently has a controlling stake within the group’s company arm.

Musk’s go well with in opposition to OpenAI accuses the startup of abandoning its non-profit mission, which aimed to make sure its AI analysis advantages all humanity. Musk had sought a preliminary injunction to halt OpenAI’s conversion. A federal choose denied the request, however permitted the case to go to a jury trial in spring 2026.

In response to the ex-OpenAI workers’ temporary, OpenAI’s current construction — a nonprofit controlling a gaggle of different subsidiaries — is a “essential half” of its general technique and “vital” to the group’s mission. Restructuring that removes the nonprofit’s controlling position wouldn’t solely contradict OpenAI’s mission and constitution commitments, however would additionally “breach the belief of workers, donors, and different stakeholders who joined and supported the group primarily based on these commitments,” asserts the temporary.

“OpenAI dedicated to a number of key ideas for executing on [its] mission of their constitution doc,” the temporary reads. “These commitments had been taken extraordinarily critically inside the firm and had been repeatedly communicated and handled internally as being binding. The courtroom ought to acknowledge that sustaining the nonprofit’s governance is important to preserving OpenAI’s distinctive construction, which was designed to make sure that synthetic common intelligence advantages humanity slightly than serving slender monetary pursuits.”

Synthetic common intelligence, or AGI, is broadly understood to imply AI that may full any job a human can.

In response to the temporary, OpenAI usually used its construction as a recruitment software — and repeatedly assured employees that the nonprofit management was “vital” in executing its mission. The temporary recounts an OpenAI all-hands assembly towards the top of 2020 throughout which OpenAI CEO Sam Altman allegedly burdened that the nonprofits’ governance and oversight had been “paramount” in “guaranteeing that security and broad societal advantages had been prioritized over short-term monetary positive aspects.”

“In recruiting conversations with candidates, it was widespread to quote OpenAI’s distinctive governance construction as a vital differentiating issue between OpenAI and opponents reminiscent of Google or Anthropic and an essential motive they need to take into account becoming a member of the corporate,” reads the temporary. “This similar motive was additionally usually used to influence workers who had been contemplating leaving for opponents to remain at OpenAI — together with a few of us.”

The temporary warns that, ought to OpenAI be allowed to transform to a for-profit, it could be incentivized to “[cut] corners” on security work and develop highly effective AI “concentrated amongst its shareholders.” A for-profit OpenAI would have little motive to abide by the “merge and help” clause in OpenAI’s present constitution, which pledges that OpenAI will cease competing with and help any “value-aligned, safety-conscious” challenge that achieves AGI earlier than it does, asserts the temporary.

The ex-OpenAI workers, a few of whom had been analysis and coverage leaders on the firm, be a part of a rising cohort voicing sturdy opposition to OpenAI’s transition.

Earlier this week, a gaggle of organizations, together with non-profits and labor teams just like the California Teamsters, petitioned California Lawyer Basic Rob Bonta to cease OpenAI from changing into a for-profit. They claimed the corporate has “failed to guard its charitable belongings” and is actively “subverting its charitable mission to advance secure synthetic intelligence.”

Encode, a non-profit group that co-sponsored California’s ill-fated SB 1047 AI security laws, cited related considerations in an amicus temporary filed in December.

OpenAI has stated that its conversion would protect its non-profit arm and infuse it with assets to be spent on “charitable initiatives” in sectors reminiscent of healthcare, schooling, and science. In change for its controlling stake in OpenAI’s enterprise, the nonprofit would reportedly stand to reap billions of {dollars}.

“We’re really on the point of construct the best-equipped nonprofit the world has ever seen — we’re not changing it away,” the corporate wrote in a sequence of posts on X on Wednesday.

The stakes are excessive for OpenAI, which wants to finish its for-profit conversion by the top of this 12 months or subsequent or threat relinquishing a few of the capital it has raised in latest months, in keeping with studies.