On February 29, local time, Musk filed a lawsuit against OpenAI, and a lawsuit was submitted to the San Francisco Superior Court. OpenAI signed an "Establishment Agreement" when it was incorporated, which stipulates that the company should operate as a non-profit organization for the purpose of developing AGI technology for human society, and maintain the principle of open source. Based on these terms, Musk completed the start-up of OpenAI with Gregory Brockman and Sam Altman. Musk believes that the current OpenAI has violated the "Establishment Agreement", so there is one of the important causes of action in this case. 
The key to the entry into force of some of the provisions is whether OpenAI has reached the level of general artificial intelligence AGI, which will be the most important line of fire in litigation. Musk believes that GPT-4 has reached the threshold of AGI, and OpenAI and Microsoft have studied AGI in more depth. Microsoft does not have a license to acquire and use AGI technology from OpenAI, but GPT-4 is integrated into many of Microsoft's products. "OpenAI has transformed into a de facto closed-source subsidiary of Microsoft, the world's largest technology company," the lawsuit says. Under its new board of directors, it is not only developing, but actually perfecting, AGI to maximize Microsoft's profits, not for the benefit of humanity. Musk filed claims against OpenAI, including breach of contract, breach of fiduciary duty, and unfair business practices, and demanded that the company reinstate open source. OpenAI, Brockman and Altman are all defendants in the case, and Musk is asking for an injunction to stop the defendants, as well as Microsoft, from profiting from OpenAI's general AI technology. For general, compensatory, and punitive damages to be determined after the trial, Musk will donate to non-profit organizations or charities. In addition, the lawsuit describes the process before and after the establishment of OpenAI, especially Musk's years of experience in founding and operating OpenAI, and outlines the context of OpenAI's development to the present. This article is based on the lawsuit, in which the events and context represent only the statements of the plaintiff.
The threat of AGI
The indictment is placed in the context of a paradigm shift in AI development and explains the origins of Musk's concerns about AI and his long-standing efforts. Before Google acquired Deepmind, Musk met Demis Hassabis, the co-founder of Deepmind. In 2012, at a meeting at the SpaceX factory, Hassabis highlighted the potential dangers of artificial intelligence. After this conversation, Musk is increasingly focusing on artificial intelligence. At the end of 2013, Musk learned that Google was planning to buy Deepmind, and he was deeply concerned about this. Because in a conversation with Google's then-CEO Larry Page, Page was indifferent to the dangers of artificial intelligence. To prevent Deepmind's powerful technology from falling into Google's hands, Musk and PayPal co-founder Luke Nosek plan to raise funds to acquire Deepmind. The two tried to convince Hassabis in an hour-long meeting, but failed. In 2014, Google acquired Deepmind, which focuses on deep learning. One of Deepmind's first development projects was Alphazero, a chess algorithm. AlphaZero knew nothing other than the rules of the game, but after being trained in "reinforcement learning", it became the most powerful chess system in the world at the time. With the help of the DeepMind team, Google quickly became the frontrunner in the AGI race. Musk is deeply disturbed by this. He believes that in the hands of a closed for-profit company like Google, AGI poses a serious danger to humanity. According to the lawsuit, in 2014, it was already difficult to compete with Google in its core business. The lawsuit emphasizes that this does not stop Musk from continuing to ensure the safe development of artificial intelligence. In 2015, Musk discussed AI and AI safety with Obama, during which Musk explained the dangers of AI and advocated for regulation. "Obama understands the dangers of AI, but regulation never came." That same year, Musk met Altman. 
Before and after the establishment of OpenAI
Altman approached Musk and said he was equally concerned about the threat of super artificial intelligence (SMI). The lawsuit refers to Altman's exact words, arguing that "the development of super-artificial intelligence may be the greatest threat to the continued existence of humanity." I think there are other threats that are more likely to happen, but are unlikely to destroy every human in the universe, but super AI can do that. Later that year, Altman suggested to Musk that they join forces to form a non-profit artificial intelligence lab to catch up with Google in the race to AGI in the opposite way. In March 2015, at Altman's suggestion, he and Musk began to prepare an open letter to the United States, calling for the regulation of artificial intelligence and contacting influential people in the field of artificial intelligence to co-sign it. The move caught the attention of Hassabis, who contacted Musk in April 2015 to confirm the matter. Five days later, Hassabis announced the first meeting of Google's Deepmind AI Ethics Committee and invited Musk to chair it as a member of the committee. 
After the first meeting, Musk recognized that this committee was just a front to try to slow down AI regulation. Musk and Altman's open letter was sent out on October 28, 2015, and was signed by more than 11,000 people, including Stephen Hawking and Stephen Wozniak. OpenAI was incorporated on December 8, 2015, with Musk and Altman appointed as OpenAI co-chairs and Brockman appointed as CTO. Prior to registration, an Incorporation Agreement was agreed that the new laboratory would develop AGI for the benefit of humanity and would be non-profit rather than seeking to maximize shareholder profits; It will be open source, balancing only opposing security considerations, and will not have its technology closed and kept secret for commercial reasons. To reflect the founding agreement, Musk named this artificial intelligence lab "OpenAI". At the same time, this agreement is also recorded in the company's certificate of incorporation. 
According to the lawsuit, Altman, Brockman and OpenAI have reiterated on multiple occasions that Musk was the driving force behind the creation of OpenAI, contributing most of the company's funding in the first few years and advising on research directions. On top of that, Musk has recruited some of the world's leading scientists and engineers, including chief scientist Ilya Sutskever. The lawsuit emphasizes that without Musk's founding contributions and early leadership, there would be no OpenAI. At the time, Google DeepMind was constantly recruiting and offering good pay, "so hiring employees for OpenAI was a difficult task." Thanks to Musk's efforts, OpenAI poached scientist Ilya from Google. The move provoked a backlash from Google, which began to make increasingly lucrative counteroffers to OpenAI's new hires. At this time, Musk believes that the salaries of existing employees need to be reconsidered, "Either we find the best people in the world, or we will be defeated by Deepmind." As long as I can cultivate an ace genius, I don't care. Deepmind puts a lot of mental pressure on me, and if they win, it will be really bad news. They are clearly making significant progress, and given the level of talent there, it is a matter of course. According to the lawsuit, Musk donated more than $15 million to OpenAI in 2016 and nearly $20 million in 2017, both times more than any other donor. Between 2016 and September 2020, Musk donated more than $44 million to OpenAI. In addition, Musk also leased OpenAI's office space in San Francisco's Pioneer Building through Musk Industries and paid a monthly rent. During this time, Musk visited OpenAI regularly and attended important milestone events, such as donating the first DGX-1 AI supercomputer to OpenAI in 2016. In 2017, Brockman et al. proposed transforming OpenAI from a non-profit organization to a for-profit organization. After a few weeks of communication, Musk said he was only funding a nonprofit and not giving free money to startups. The communication ended with Brockman et al. committing to continue maintaining OpenAI's non-profit structure. On February 21, 2018, Musk resigned as co-chairman of OpenAI, but still donated about $3.5 million that year. On March 11, 2019, OpenAI announced that it would create a for-profit subsidiary, OpenAI LP, and after the news was announced, Musk contacted Altman and asked him to "make it clear that I have no financial interest in OpenAI's for-profit division." That year, Musk donated an additional $3.48 million.
OpenAI went from open source to closed source
OpenAI's initial work largely followed in the footsteps of Deepmind, using reinforcement learning to play games. However, instead of playing chess, OpenAI plays a strategy game, Dota 2. The OpenAI team built a model that defeated the defending world champion team, demonstrating how "self-playing" reinforcement learning can achieve superhuman performance on difficult tasks. At the same time, Google has created an algorithm called "transformers" to solve many of the problems that deep learning faces when it comes to understanding long text sequences. OpenAI's researchers found that the Transformer algorithm can perform many natural language tasks without pre-training. What's even more surprising is that by using the Transformer architecture, a pre-trained deep neural network can generate new text. In January 2018, OpenAI released the source of GPT and the training model, as well as a detailed article describing the model and its functions. In 2019, OpenAI released GPT-2, which also has a detailed ** describing this model. In conjunction with the release of GPT-2, OpenAI also released a detailed ** co-authored by OpenAI scientists and independent social and technical scientists, explaining the many benefits of publicly releasing the model instead of a closed one. In 2020, OpenAI released the third version of its model, GPT-3, while announcing that it would make public the full study** for others to learn from the model. GPT-3 opened a path to generative AI, but that path was dramatically reversed, according to the lawsuit. On September 22, 2020, OpenAI reached an agreement with Microsoft to exclusively license its GPT-3 model to Microsoft. However, Microsoft's license only applies to OpenAI's pre-AGI (pre-AGI) technology, and Microsoft does not acquire any rights to AGI. On March 14, 2023, OpenAI released a new generation model, GPT-4. GPT-4 is a completely closed model, and OpenAI has not published anything describing its internal design**. The release of GPT-4 is a sign that, according to the lawsuit, Musk argues that Altman has made OpenAI fundamentally deviate from its original mission and historical practice of making its technology and knowledge available to the public. The interior of GPT-4 is designed to be kept completely secret, "primarily for commercial reasons rather than security reasons." In addition, Musk, as the plaintiff, argues that GPT-4 is an AGI algorithm and is determined not to fall within the scope of the exclusive license that Microsoft signed with OpenAI in September 2020. But "GPT-4 is now actually Microsoft's proprietary algorithm, which has been integrated into the Office software suite". The lawsuit cites a Microsoft researcher as publicly stating that "given the breadth and depth of GPT-4's capabilities, we reasonably believe that it can be considered an early version of a general AI system." Whether GPT-4 has reached the threshold of AGI will be key in this lawsuit.
Let's mention the farce of "palace fighting" again
Since Microsoft only has rights to some of OpenAI's pre-AGI technologies, the board's decision on whether OpenAI will reach AGI levels will be key to the partnership. But "in a series of shocking developments, most of OpenAI's board of directors were forced to resign on November 22, 2023, and their replacements were handpicked by Altman and Microsoft," the lawsuit said. On November 17, 2023, OpenAI's board of directors fired Altman because "he did not consistently and honestly face the board" and "lost confidence in his ability to continue to lead OpenAI." According to the lawsuit, in the next few days, Altman, Brockman and Microsoft joined forces to use Microsoft's huge influence on OpenAI to force the resignation of most members of OpenAI's board of directors. Altman returned to OpenAI on November 21, and the new board members were hand-picked by Altman and supported by Microsoft. "OpenAI's once-elaborate nonprofit structure has been replaced by a purely profit-oriented CEO and a board of directors with a low level of expertise in AGI and AL public policy," the lawsuit reads, "The new board members lack AI expertise and simply do not have the ability to independently judge whether and when OpenAI will reach AGI," the lawsuit reads. Therefore, it does not have the ability to independently judge whether the algorithm developed by OpenAI is beyond the scope of the license to cooperate with Microsoft." The lawsuit cites news reports that Altman's dismissal was due in part to OpenAI's breakthrough in AGI. OpenAI's board members and executives are divided over security issues and the potential threat posed by Q-Star, the next-generation algorithm. According to reports, several OpenAI staff members have written a warning letter about the potential power of Q-Star, which could be a clearer illustration of OpenAI's development of AGI. The lawsuit mentions two other people in the farce, Microsoft CEO Satya Nadella and Helen Toner, a former board member of OpenAI. According to the lawsuit, Microsoft believes that through its substantial ownership of OpenAI, it could completely deprive OpenAI of its research if it ceases to exist. In an interview shortly after Mr. Altman's dismissal, Mr. Nadella said: "We are very confident in our abilities. We own all the intellectual property rights and all the capabilities. ”
According to the lawsuit, Microsoft never gave up on its plan to ensure that Altman was reinstated in his position, and in the days after Altman's dismissal, OpenAI's board of directors faced increasing pressure from lawyers and major shareholders, including Microsoft, who were asked to reinstate Altman. The complaint mentions that Ms. Tonna was a particular target for the reinstatement of Altman. During this time, an OpenAI lawyer told Tonna that if OpenAI went bankrupt as a result of Altman's firing, she and the board could face claims because they have a fiduciary duty to investors. Tonna described the lawyer's warning as a tactic of intimidation because OpenAI's board of directors has never had a fiduciary duty to investors, even its for-profit companies, and all investors were told that the company's responsibility for its mission took precedence over its responsibility to its investors. Microsoft had promised a $10 billion investment in OpenAI, and when the farce unfolded, Microsoft paid only a fraction. This gives Microsoft a huge deal of influence over the "independent" nonprofit board. In addition, if Microsoft refuses to provide the cloud computing system that OpenAI relies on, OpenAI will be incapacitated. After these events in 2023, OpenAI now essentially completely upends the agreement it had when the company was founded, the lawsuit concluded. OpenAI has turned into a closed-source subsidiary of Microsoft, the world's largest technology company. Under the new board of directors, the company is not only developing, but actually improving AGI to maximize Microsoft's profits. It is worth noting that this indictment follows a report by Wall Street that the U.S. Securities and Exchange Commission (SEC) was reviewing the internal communications of OpenAI's current and former, directors, including Altman, and issued a subpoena to OpenAI in December last year. The investigation is aimed at OpenAI's "palace fight" farce, which the SEC believes is suspected of misleading investors. If Musk's case against OpenAI is heard, it is foreseeable that more details of this incident will be made public. In addition, the lawsuit asserts that the judicial determination of GPT-4 constitutes artificial general intelligence, which will become the focus of contention between the two parties, and in this process, OpenAI and GPT-4 will inevitably be unveiled.