Elon Musk Sues OpenAI and Sam Altman for Breach of Contract
Elon Musk has recently filed a lawsuit against OpenAI, alleging that CEO Sam Altman and other executives had departed from the startup's initial mission and, in doing so, “breached the founding agreement”.
Musk, a co-founder no longer actively involved with OpenAI, asserts that the company's focus on profit contradicts its original commitment to benefit humanity, as stated in its charter on OpenAI's webpage. The legal action, lodged in San Francisco, claims that OpenAI's close ties with Microsoft have compromised its founding mission, particularly in terms of creating open-source technology free from corporate influence. Musk contends that "OpenAI Inc. has been transformed into a closed-source de facto subsidiary of the largest technology company in the world: Microsoft."
Microsoft, as one of the leading entities in artificial intelligence, recently surpassed Apple as one of the world's most valuable companies, now boasting a market value of $2.89 trillion. This success is attributed to its significant stake in OpenAI and the successful integration of AI solutions across its business. An analysis by Paul Allison at Finimize considers the potential productivity savings using a "top-down" analysis for knowledge workers using Microsoft's AI-infused software. Assuming this software saves two hours daily for knowledge workers, each earning an average of $40 per hour, the estimated savings per worker amount to $80 per day. If Microsoft retains $5 of this $80 through increased prices on its AI-loaded software, it could result in $6 per Microsoft user per day, potentially generating $546 billion in revenue annually.
Amid a broader global competition inquiry into Microsoft's OpenAI partnership, an alteration on OpenAI's website, changing Microsoft's status from "minority owner" to "minority economic interest", attracted attention. Microsoft's $13 billion investment entitles it to a potential 49% of subsidiary profits, structured under a "capped profit" system. The majority ownership lies with a holding company managed by OpenAI's stakeholders, avoiding direct equity. Exclusive rights to OpenAI's existing IP, excluding future AGI-related IP, are held by Microsoft. The OpenAI board retains the authority to declare AGI breakthroughs.
It is essential to outline Elon Musk's involvement with OpenAI. Musk co-founded OpenAI in 2015, establishing it as a non-profit organisation aimed at addressing potential threats posed by artificial intelligence. Concerns arose when Google acquired DeepMind in 2014, prompting Musk's failed attempt to buy it. Greg Brockman, a member of OpenAI's board, proposed transitioning the company's structure to a for-profit model, allowing it to secure funding for the substantial computing power required for language model development. Musk opposed this idea, expressing in an email to Brockman, Ilya Sutskever, and Altman, that they should either pursue a for-profit model independently or continue with OpenAI as a non-profit.
In response to this claim by Musk, OpenAI stated that Musk acknowledged the company's need to pursue profitability but sought majority equity, initial board control, and the role of CEO. Reports surfaced in late 2017 suggesting that Musk aimed to take control of OpenAI from Altman and the other founders, intending to transform it into a commercial entity in collaboration with Tesla, utilising the automaker's supercomputers. Altman and others resisted this move, leading to Musk's resignation. He cited a desire to focus on Tesla's AI projects and officially announced his departure during a meeting with OpenAI staff in February 2018.
Musk's lawsuit includes claims of breach of contract, breach of fiduciary duty, and unfair business practices. The primary contention is the introduction of GPT-4, asserted as Artificial General Intelligence (AGI). AGI can be seen at which computers function at or above the level of human intelligence. Musk seeks a ruling to categorise GPT-4 and the advanced technology Q* as AGI, thereby excluding them from Microsoft's OpenAI license. The Microsoft deal only gives the tech giant a licence to OpenAI’s pre-AGI technology, hence this lawsuit hinges on defining the threshold for AGI. The lawsuit points to occasions where GPT-4 reportedly surpasses human performance, such as scoring above the 90th percentile in the bar exam and achieving results within the 99th percentile in the Biology Olympiad for high school students. OpenAI maintains it hasn't achieved AGI despite GPT-4's success, as they are vulnerable to generating errors, fabrications and “hallucinations”. Strangely, Musk has publicly expressed the opposite, stating on X in December 7, 2022, "To be called AGI, it needs to invent amazing things or discover deeper physics - many humans have done so. I haven't seen that potential yet," suggesting that, by his own criteria and perception of AGI, GPT-4 has not achieved these feats.
The initial criteria for OpenAI as a non-profit AGI developer included being not-for-profit and open-source. Public lawsuits, like the New York Times suing OpenAI for using millions of reports as training material without permission, have raised concerns about GPT-4's secretive nature. The lawsuit also highlights OpenAI's dramatic restructuring in 2023, emphasising Altman's firing and subsequent reinstatement. Musk alleges this restructuring, coupled with Microsoft's leverage, aimed to hide the existence of AGI from the new board. The claim suggests intentional obfuscation to prevent Microsoft from licensing AGI if already achieved.
Moreover, in the lawsuit, Musk seeks for a court order to be issued that requires OpenAI to make all its research and technology public, and Altman to relinquish any earnings obtained from alleged unlawful practices. Musk also pursues unspecified damages, intending to contribute any won compensation to charity.
Legal expert Mark Lemley from Stanford Law School expresses skepticism about the AGI claim and questions the lawsuit's broader legal foundation. Lemley highlights the absence of a documented contract between Musk and the company, emphasizing the need for contractual specifics to support the legal arguments. Musk's reliance on "promissory estoppel," a legal principle preventing parties from reneging on promises causing reliance-based harm, raises concerns about the inadequacy of establishing a formal contract. Ann Lipton, a business law professor at Tulane University, emphasises that while a non-profit company is limited in distributing profits to equity holders, there is no prohibition on forming for-profit subsidiaries, allowing non-profits to engage in commercial ventures through subsidiaries while maintaining the charitable or non-distributive nature of the parent organisation.
There are many covert and overt motivations driving the legal saga involving Elon Musk, OpenAI, and Microsoft. While some motivations, such as safety concerns, have been explicitly stated, the call for transparency and public disclosure in OpenAI's practices underscores the broader debate surrounding responsible AI development. However, beneath the surface, business interests lie, as exemplifed by Musk's xAI company which emerges as a direct competitor to OpenAI. The legal proceedings are mere embellishment for what can be perceived to be personal betrayal at the heart of this dispute, from Musk investing $44 million into OpenAI operations from 2016-2022 to now painting Altman’s approach to AI development as “a stark betrayal of the founding agreement.” The outcome of Musk's lawsuit against OpenAI and the ongoing investigations into Microsoft's involvement will likely have far-reaching implications, shaping the trajectory of AI development and its governance in the years to come.
By Sumaiyah Patel
