OpenAI co-founders questioned Altman's motives back in 2017

If you thought OpenAI’s palace coup in late 2023 was a sudden bolt from the blue, newly unsealed court records just dropped a reality check.

Emails released as part of Elon Musk’s ongoing lawsuit against OpenAI show that the company’s internal drama, power struggles, and deep paranoia date all the way back to 2017.

Back then, the cofounders were wrestling with a core dilemma: who should run the ship, and should the project ditch its non-profit roots for a for-profit structure?

The original mistrust

On September 20, 2017, Greg Brockman and Ilya Sutskever sent a blistering joint email to Sam Altman.

The two technical leaders stated plainly that they had not been able to fully trust Altman's judgment throughout the reorganization process.

Then came the pointed questions: Is AGI truly your primary motivation? And how does this connect to your political goals?

Musk didn’t escape scrutiny either. In another note, Brockman and Sutskever called out Musk's hunger for dominion, writing that while he claimed he didn't want to control the final AGI, his negotiations showed that absolute control was paramount to him. Creating a setup where you could become a dictator is simply a bad idea, they warned.

Neither leader took the pushback well. Shivon Zillis, an adviser and close Musk confidante, noted Altman admitted he had lost significant trust with Greg and Ilya. Musk declared the pushback the final straw, telling them to either go do something on their own or keep OpenAI strictly as a non-profit.

DeepMind panic, Microsoft insults, and a crypto pitch

The stash also reveals fascinating behind-the-scenes chaos from the early days of the AI race.

Musk was openly stressed about Google’s DeepMind, writing in 2016 that the rival lab gave him extreme mental stress and warning that Demis Hassabis could establish an AGI dictatorship with a one mind to rule the world mindset.

Altman shared similar fears. In 2015, he noted that stopping AI progress was likely impossible, concluding that it would be better for someone other than Google to reach the finish line first. Altman even claimed DeepMind once prepped massive counteroffers to poach the entire OpenAI staff and kill the startup.

Meanwhile, Musk had visceral reactions to early corporate deals. Reviewing a proposed 2016 partnership with Microsoft, Musk bluntly wrote that the terms made him feel nauseous, adding it would be worth far more than $50 million not to look like Microsoft's marketing puppet.

To raise money, Altman even floated the idea of launching an Initial Coin Offering (ICO). Musk shot it down immediately, warning that launching a crypto token would destroy the organization’s credibility.

Non-profit backers were equally uneasy. Open Philanthropy, which had awarded a $30 million grant in 2017, resisted the commercial pivot. Its head, Holden Karnofsky, suggested more funding if OpenAI stayed pure, prompting Altman to float an ambitious $100 million price tag.

Why it matters

These emails completely demolish the clean, idealistic mythos often spun around the creation of frontier AI.

For developers and founders relying on OpenAI's infrastructure today, this paper trail provides crucial context. The board clash that briefly ousted Altman was not a bizarre, one-off anomaly. The exact same fault lines—fears of Altman’s personal ambitions and doubts about governance—were explicitly documented six years earlier.

In our view, this history shows just how rapidly survival and raw compute costs can dismantle founding principles. Musk once claimed taking Microsoft’s cash would turn them into puppets; today, OpenAI is inextricably bound to Azure. Altman once explored dubious crypto tokens; today, he runs a multi-billion-dollar commercial powerhouse.

It appears that building AGI is fundamentally impossible to insulate from power politics and Wall Street realities. The takeaway for the broader industry is clear: never base your business strategy on the presumed benevolence or philosophical mission of an AI provider. When push comes to shove, survival and control will always override founding ideals.