From: allin
Elon Musk has filed a lawsuit against OpenAI, its CEO Sam Altman, and President Greg Brockman, alleging breach of contract, breach of fiduciary duty, and unfair competition [01:31:00]. The lawsuit was filed shortly after the recording of a previous podcast episode, sparking widespread discussion and “meme Wars” [01:21:00].
Background and Allegations
OpenAI was initially established as an open-source, non-profit organization [01:41:00]. Elon Musk contributed between 75 million to the organization, with the understanding that its technology would remain open source for the benefit of all [01:45:00]. The core concern for Elon Musk was to prevent any single large tech company, initially Google and now Microsoft, from controlling AI [03:39:00].
However, OpenAI transitioned into a closed-source, for-profit venture, forming “OpenAI Global LLC” in 2019 while retaining a non-profit entity [01:51:00]. This change allowed them to benefit from non-profit tax breaks during development before moving to a for-profit model [02:02:00].
Elon Musk contends he was “swindled” or “cheated” as the company’s fundamental purpose was altered [04:00:00]. He argues that if this model is allowed, it could set a precedent for other companies to start as non-profits, leverage pre-tax dollars, and then convert to for-profit entities, circumventing taxes [04:17:00]. He seeks for OpenAI to open source its models and for shareholders to receive no financial benefit [02:32:00].
Additionally, Elon Musk claims OpenAI aggressively used his name to raise funds and that he was instrumental in recruiting early critical hires, including Ilya Sutskever, from Google [06:07:00].
OpenAI’s Defense and Structural Nuances
OpenAI has responded by releasing Elon Musk’s emails, suggesting he acknowledged the necessity of a for-profit subsidiary to raise billions of dollars to compete with entities like Google DeepMind [04:34:00] [09:02:00].
David Friedberg suggests that a “market-based system” with for-profit incentives was needed to secure the capital required for advanced AI development, drawing a parallel to the Cystic Fibrosis Foundation’s investment in Aurora Biosciences [10:00:00]. The foundation invested 3.3 billion, demonstrating “venture philanthropy” [10:11:00].
The key legal questions for OpenAI’s non-profit status are whether the non-profit entity still owns a meaningful piece of the for-profit and if it continues to perform “charitable stuff” [11:57:00]. Concerns arose when “non-profit experts” were removed from the board, potentially crossing a “tripwire” [12:44:00].
Chamath Palihapitiya argues that while OpenAI’s legal structure allows for non-profit investment in for-profit entities, the transfer of IP and employees from the non-profit to the for-profit, coupled with significant equity grants and secondary sales to employees, is problematic [13:26:00]. He compares it to the Mozilla Foundation, which set up separate boards and ensured all profits flowed back to the non-profit, with employees receiving only above-market salaries, not equity [14:30:00]. The IRS investigated Mozilla for tax reasons despite these precautions [15:22:00].
OpenAI’s structure involves a non-profit, a holding company, and the for-profit OpenAI Global LLC, with Microsoft reportedly owning 49% of the for-profit [15:07:00]. The remaining 51% is held by a combination of the non-profit, employees, and investors [15:32:00].
The OpenAI LP agreement states that the partnership’s duty to the mission (ensuring safe AGI benefits humanity) takes precedence over generating profit [34:30:00]. It even allows the general partner to reinvest all cash flow into R&D without obligation to limited partners, implying investors could potentially lose their money [34:46:00].
The Open Source vs. Closed Source Debate
A central point of contention is OpenAI’s shift from an open-source model, which was meant to allow humanity to benefit from the code, to a closed-source model [20:16:00]. Critics argue this change allows employees and investors to capture value that was intended for broader public benefit [24:20:00]. Some employees have reportedly sold $2 billion in secondary shares [24:39:00].
While Apple and Meta are producing open-source AI software, OpenAI claims its code is “too dangerous” to be open source [26:47:00].
Future Outlook and Broader Implications
The lawsuit highlights a potential loophole in the US tax system that could motivate many companies [06:47:00]. The IRS is expected to scrutinize the case closely due to the significant money involved and the wealth transfer to employees [15:10:00] [25:16:00].
If OpenAI were to revert to an open-source model, Elon Musk has implied he would drop the lawsuit [05:36:00]. However, this wouldn’t resolve the broader tax and structuring issues now “cracked open” for the entire non-profit industry [31:07:00].
There’s speculation that if OpenAI had been a for-profit from the start, Elon Musk’s initial investment could have given him significant ownership, possibly 20% or more, worth billions today [32:02:00].
Sam Altman’s “OpenAI Ventures fund,” which he was initially the sole owner of (though now claimed to be a clerical error), raises “corporate opportunity issues” where the economic opportunity should belong to OpenAI itself [21:14:00]. This convoluted compensation structure, where he has “retained optionality to monetize the ecosystem” rather than direct equity, is seen as problematic [32:53:00]. The SEC is reportedly looking into these matters [34:09:00].
The Pursuit of AGI
OpenAI’s stated mission is to create Artificial General Intelligence (AGI) [38:50:00]. The definition of AGI is debated, ranging from AI that can replace 80% of jobs to being smarter than the smartest human being [39:10:00] [41:56:00]. Critics express concern about the “cultish and weird” aspect of explicitly pursuing AGI, which in common parlance can mean “Skynet” or a “sentient God” [41:31:00] [41:16:00].
Proponents, however, view AGI as a tool that could profoundly expand humanity’s capabilities, enabling individuals to accomplish tasks that would otherwise require massive collective effort, such as designing a Mars mission or curing cancer [39:39:00]. The recent “Claude 3 Model” by Anthropic, which demonstrated the ability to solve complex quantum physics problems and complete screenplays, showcases the impressive capabilities of current AI advancements [42:24:00].