Is Musk suing OpenAI for the well-being of humanity?

Written by: Mu Mu

Elon Musk has confronted OpenAI more than once, and this time he has taken a significant step by suing OpenAI in the U.S. District Court in San Francisco.

On March 1, reports indicated that Musk accused OpenAI and its CEO Sam Altman of violating the agreement made at its inception, which stated that technology should be developed for the benefit of humanity rather than for profit. The demand is for the court to issue an injunction prohibiting OpenAI and Microsoft from profiting from the company's General Artificial Intelligence (AGI) technology.

Musk's renewed attack on OpenAI's "argument" still revolves around the risks of artificial intelligence, once again throwing "consideration for human welfare" into the public discourse.

Can a leader of a commercial company genuinely consider the interests of humanity? As the public wonders, OpenAI updated its blog stating, "We intend to take action to dismiss all of Elon’s claims," and even released email exchanges between Musk and OpenAI's founding members, proving that he once wanted to merge it with Tesla or take control of OpenAI. Disagreements arose, leading Elon Musk to leave OpenAI.

The "Iron Man" lawsuit against the AI giant has unearthed old tales of "Game of Thrones," while the entanglement between the two also hints at the future of AI competition.

Musk Raises the Banner of "Human Welfare" Against OpenAI

"Microsoft" and "closed source" are key terms in Musk's lawsuit, which also serve as evidence for his accusation that OpenAI has deviated from its original intention of "developing artificial intelligence technology for the welfare of all humanity rather than for profit."

In the lawsuit, Musk claims that OpenAI's relationship with Microsoft in recent years has severely strayed from the organization's goal of developing public open-source AGI, and that OpenAI has effectively become a closed-source subsidiary of the world's largest tech company, Microsoft.

Musk also cited that the GPT-4 released in March 2023 is still a closed-source model compared to previous versions, with the closed nature being driven by commercial considerations rather than human welfare.

Whether GPT-4 has reached the threshold of AGI will be a key point in this lawsuit. Besides GPT-4, Musk also mentioned in the lawsuit that "it is reported that OpenAI is currently developing a model called Q*, which is more likely to be AGI."

"Q*" was first mentioned during the incident in November last year when OpenAI CEO Sam Altman was "ousted."

On November 23, Reuters cited sources saying that four days before Altman was ousted, several researchers had warned the board about a powerful AI discovery within OpenAI that could threaten humanity. Subsequently, insiders stated that OpenAI acknowledged in an internal letter to employees that there was a project named Q*, which some believed could be a breakthrough in the company's exploration of AGI.

Thus, Q* circulated in the industry as an "OpenAI secret project," and the outside world viewed it as AGI technology that OpenAI was developing.

Coincidentally, after Musk sued OpenAI, a 53-page PDF document claiming that OpenAI "will achieve AGI by 2027" went viral on the internet. This document also mentioned the mysterious Q* and stated that its next phase "was initially GPT-6 but has been renamed GPT-7," and that "the original plan was to release it in 2026."

However, netizens discovered that this timely leaked PDF came from an X account registered in July 2023, and the identity of the publisher @vancouver1717 is unknown, as they had only posted two tweets. Therefore, supporters of OpenAI expressed doubts about the authenticity of this PDF.

The closed-source GPT-4, the mysterious Q*, along with the largest profit-making entity Microsoft, together constitute the evidence Musk uses to accuse OpenAI of not considering human welfare.

Altman tweeted after Musk's lawsuit

After Musk's lawsuit, OpenAI CEO Sam Altman tweeted on X, "The hurricane is spinning faster, but it is completely calm in the eye of the storm," seemingly revealing his calm mindset in response to the lawsuit.

Perhaps Sam and OpenAI have become accustomed to Musk's criticisms.

Since the end of 2022, when OpenAI's ChatGPT sparked a global storm in natural language models, Musk, a former founding member, has appeared unusually anxious, repeatedly accusing OpenAI of not being open (open-source) and ChatGPT of being "unsafe," while also directing his criticism at OpenAI's largest investor, Microsoft.

In April last year, Musk accused Microsoft of illegally using Twitter data for AI training and stated, "Now is the time to sue." After acquiring Twitter and renaming the social platform X, Musk made a series of adjustments to X's API to prevent other AI companies from accessing the data. Meanwhile, Microsoft's social media advertising management service Smart Campaigns announced it would no longer support X.

The grievances between Musk and OpenAI date back to 2015 and erupted after OpenAI became successful. This lawsuit appears to be a rehash of old grievances. However, this time, OpenAI seems unwilling to remain silent.

OpenAI Releases Emails Exposing Musk's "Self-Interest"

On March 5, OpenAI updated its official blog regarding its relationship with Elon Musk, reiterating its mission to "ensure AGI benefits all of humanity" and clearly stating, "We intend to take action to dismiss all of Elon’s claims."

More strikingly, the blog detailed the years of entanglement between Musk and OpenAI while releasing email exchanges that proved Musk's previous "calculations" regarding OpenAI—merging with Tesla or taking control of OpenAI.

In the external version of OpenAI's founding, Musk, as a co-founder, planned to donate $1 billion to OpenAI in 2015 to support the organization in developing artificial intelligence as a non-profit entity. After donating $100 million, he had a "difference in ideology," leading to his departure, and OpenAI subsequently transitioned to a partially profit-making structure, which allowed for Microsoft's investment.

However, the blog updated on March 5 revealed more details that contradicted the external version.

The blog post disclosed that as a non-profit organization, OpenAI actually raised less than $45 million from Musk, while raising over $90 million from other donors. Moreover, at the end of 2015, when Greg and Sam founded OpenAI, the initial funding goal was only $100 million, but Musk suggested "a much larger number than $100 million," stating, "I think we should say we will start with a commitment of $1 billion in funding… I will cover the funds that others do not provide."

By early 2017, as the team realized that building AGI required massive computational power, OpenAI recognized the need for billions of dollars annually, far beyond what a non-profit organization could raise, which forced the OpenAI founders, including Musk, to discuss a profit-oriented structure.

During this process, "Elon wanted us to merge with Tesla, or else he wanted to have complete control over us." OpenAI stated in the blog that in 2017, when deciding to create a profit-making entity, "Elon wanted to obtain a majority stake, preliminary control of the board, and serve as CEO. During these discussions, he withheld funding." Subsequently, the two sides could not reach an agreement on profit-sharing terms, "because we felt that any individual having absolute control over OpenAI contradicted the company's mission."

Musk once suggested that OpenAI should rely on Tesla for funding

After that, Musk suggested merging OpenAI with Tesla. "In early February 2018, Elon forwarded us an email suggesting that OpenAI should 'treat Tesla as its cash cow,' and commented, 'Absolutely correct… Tesla is the only path that has a chance of competing with Google.' Even so, the possibility of competing with Google was still very small, but at least it was not zero."

However, Musk chose to leave OpenAI, believing its success rate was zero, and planned to establish a competing company within Tesla, stating at the end of February 2018 that he supported the team in finding their own path to raise billions of dollars. By December 2018, Musk emailed stating, "Even raising hundreds of millions is not enough. It requires billions of dollars annually, or forget it." The collaboration between the two sides completely fell apart.

Regarding Musk's accusation of "closed source" in the lawsuit, OpenAI stated in the blog that Musk knew AGI was not in a hurry to be open-sourced. The blog released emails between OpenAI scientist Ilya and Musk, stating, "As we get closer to building artificial intelligence, it makes sense to be less open. Open AI means that everyone should benefit from the results of AI, but not sharing science is perfectly fine… Elon responded, 'Yes.'"

At the end of the blog post, OpenAI expressed regret, "All of this happened to someone we deeply admire—he inspired us to pursue higher goals and then told us we would fail, creating a competitor, and when we began to make meaningful progress towards OpenAI's mission without him, he sued us."

Musk's Struggles in Artificial Intelligence

OpenAI's blog post, accompanied by email screenshots, elaborated on Musk's past "ambitions" for OpenAI, which also undermined his accusations of "closed source."

This "Game of Thrones" that once occurred in OpenAI's early days may be one of the reasons Musk harbors resentment towards this AI giant. As we enter the era of explosive growth in natural language models, Musk, who aims to transform Tesla into a smart robotics company, must now face competition in the field of artificial intelligence after missing out on OpenAI.

After signing a joint letter to halt the training of more advanced GPT models last year, Musk expressed his intention to develop a safe and human-friendly generative dialogue product called TruthGPT.

In November last year, Musk's company xAI launched a chatbot product called Grok, directly competing with OpenAI's ChatGPT. Although Musk personally promoted Grok on social media, its popularity over the past year has not matched that of ChatGPT.

Comparison of Google search indices for Grok and ChatGPT

Now, in the market for large AI models, there are not only Google and OpenAI but also Anthropic, the creator of Claude3, increasing the competitive pressure on xAI.

Musk also intends to integrate xAI with his existing businesses. On xAI's official website, the company explicitly states that it will work closely with X and Tesla. It is important to note that Musk's xAI does not have an equity relationship with X or Tesla.

Tesla shareholders seem unprepared for the company's shift towards smart robotics. Shareholders have stated that unless Musk holds at least 25% of the voting rights in Tesla, he has no authority to develop the car manufacturer into a leader in artificial intelligence and robotics.

Musk's pursuit of artificial intelligence has not been smooth sailing; the anxious "bullet" ultimately turned towards his 2015 self.

According to CNBC, OpenAI Chief Strategy Officer Jason Kwon sent a memo to its employees in response to Musk's lawsuit, stating, "We believe the reason behind the lawsuit may be that Elon regrets not continuing to participate in the company's development. It is very disappointing to see Elon take such action against a company he helped found, especially considering he closely collaborated with those of you who are still fighting for this mission."

However, Musk's lawsuit may serve other purposes. Regarding the open-sourcing of GPT-4 and its subsequent versions, as well as further disclosures about Q*, well-known AI blogger Rowan Cheung believes this lawsuit could help unravel these mysteries.