Musk Wiki

The OpenAI Arc

NextThe Political Turn

The OpenAI Arc

In 2015 Musk helped found OpenAI. By 2023 he had turned on it, and on the man running it. The story in between is the most-repeated one in his AI thinking. He starts as co-founder and democratization evangelist in 2015–2016, becomes an adversary by 2023, and the wreckage of that fallout becomes the reason he builds xAI. The thread touches his fear of concentrated AI, his break with Sam Altman, and the founding of xAI.

Summary

The shape is clean. The same fear of concentrated AI that worried Musk also handed him his first fix. In 2015 he co-founds OpenAI on a single bet: spread the technology so no one actor can own it. In 2016 the tone is warm and collaborative.

Then it sours. By 2018–2020 the first public cracks show: “OpenAI should be more open”, plus a named distrust of its safety leadership. In 2023 the break is total. OpenAI has gone closed and for-profit (“not good karma,” “super closed source for maximum profit”). And during the November 2023 board crisis, Musk reads its CEO Sam Altman through the image of power that can corrupt (“ring of power”). But the break isn’t an ending. The same grievance becomes the reason he builds xAI. He casts it as a truth-seeking lab, carrying forward the “opposite of an unsafe Google” logic he first aimed at OpenAI’s own founding.

So the institution reversed while the conviction held. The instinct to open-source as a hedge never moves. What moves is that the vehicle Musk built to carry it became, on his telling, the exact thing he founded it to prevent. So he built another.

2015–2016 — co-founding OpenAI as the democratization remedy

He starts as a builder, not an enemy. By 2016 Musk already ranks AI as the single biggest near-term risk to humanity, the worry traced in full on AI existential risk. His fix is to distribute the technology so no one company or person controls it. In the September 2016 Y Combinator conversation, interviewed by his own OpenAI co-founder Sam Altman, he names that fix outright:

“is that we achieve democratization of AI technology, meaning that no one company or small set of individuals has control over advanced AI technology.” ↗

And the reason he gives for co-founding OpenAI is exactly this: distribute the technology to minimize existential risk. The tone is warm and mission-driven:

“I think people really believe in the mission. I think it’s important and it’s about minimizing the risk of existential harm in the future.” ↗

This is the arc before anything goes wrong. OpenAI is the good answer, the embodiment of democratized AI that holds the existential risk down, and Altman is a partner, not an opponent.

2018–2020 — the first cracks

The first public fractures show up in the 2018–2020 tweets, two to four years after the warm 2016 talk. The complaint is, fittingly, that the open organization has stopped being open:

“OpenAI should be more open imo” ↗

In the same February 2020 exchange he goes further. He names a loss of influence and a specific distrust of its safety leadership, the seed of the years-long feud:

“I have no control & only very limited insight into OpenAI. Confidence in Dario for safety is not high.” ↗

These two tweets are where the rupture begins: the openness critique and the named safety doubt that grow into the full 2023 break. But the grievance is still about the organization’s direction. It isn’t yet a verdict on a person.

2023 — the full rupture

By 2023 the cracks become a clean break, and he says so in three separate sittings. In the November 2023 Lex Fridman conversation, Musk gives his most compressed verdict: the name itself has become a lie.

“the open in open AI is supposed to mean open source, and it was created as a nonprofit open source, and now it is a closed source for maximum profit, which I think is not good karma” ↗

The same month, at the DealBook Summit, days into the board crisis that briefly ousted Altman, he tells the whole arc at length. First the founding rationale the grievance rests on:

“the reason for starting OpenAI was to create a counterweight to Google and DeepMind” ↗

Then the verdict that inverts the organization’s own name:

“It is in fact a closed source, super closed. It should be renamed super closed source for maximum profit AI.” ↗

Then, asked directly about Sam Altman days into the board crisis, he gives the one direct on-the-record characterization of the man in the cited material, framed as corruption by power:

“The ring of power can corrupt, and he has the ring of power.” ↗

The 2016 partner is now the man holding the “ring of power” over an organization Musk says betrayed its open, nonprofit founding. The charge isn’t new to November either. The May 2023 CNBC interview carries the same founding-betrayal premise in its earliest 2023 form, months before the board crisis sharpened it.

2023 onward — the rupture as the founding motive for xAI

The rupture is a launch point, not an end. The same grievance, that OpenAI abandoned safety for profit, becomes the reason Musk builds xAI as a truth-seeking lab set against the labs he distrusts. It is the 2016 democratization fix tried a second time, in a new shell. And when he explains why he ever started an AI lab, he keeps reaching for the same picture he first drew at OpenAI: a thing built to be the opposite of Google. In the February 2025 Joe Rogan conversation, calling OpenAI’s reversal “an irony maximizer,” he says it again:

“I wanted to start something that was the opposite of Google because I was concerned about Google – Google wasn’t paying enough attention to AI safety, in my opinion.” ↗

That is the irony the arc turns on. The lab founded as a counterweight to Google (the DealBook line above) became, on his telling, the concentration-of-power problem itself. So he ran the same “opposite of an unsafe Google” logic one turn further out and built xAI. The truth-seeking goal baked into xAI is the same old fear in a new vehicle, after the first vehicle turned on him.

What the arc reveals

  • The conviction held; only the vehicle reversed. The open-source-as-hedge instinct runs unbroken from 2016 to 2023. What changed is that the institution Musk built to embody it became, on his telling, the concentrated-power problem it was meant to solve. It’s the clearest documented evolution of the arcs here — an evolution of circumstance, not belief.
  • The grievance narrows onto a person. In 2016 the target is an abstraction: concentrated AI. By 2018–2020 it’s the organization’s direction. By late 2023 it lands on a single man, Altman and his “ring of power.” The worry tightens from structure to institution to individual.
  • Rupture is generative for him. The break doesn’t end in resignation. It produces xAI. The pattern repeats across his life: a defeat becomes the seed of the next build. Here the existential-risk worry that started OpenAI is the same one that restarts as xAI.

Connections