[Bloomberg Column] OpenAI's Hypocrisy, Musk Is Right

Lost Initial Goals Amid Big Tech Pressure

Elon Musk, CEO of Tesla, could be called an "epic troll" (someone who deliberately teases other players). However, the lawsuit Musk filed against OpenAI on the 29th of last month offers a new perspective on the humanitarian stance of AI companies.

Musk sued AI company OpenAI and its CEO Sam Altman, alleging that Altman violated OpenAI's founding charter to build AI systems "for the benefit of humanity." Although OpenAI still displays that goal on its website, Musk calls it nonsense. In the complaint submitted to the San Francisco court, he stated, "In reality, OpenAI has transformed into a closed-source subsidiary of Microsoft (MS), the world's largest tech company."

Musk is right. But first, we must remember why Musk filed the lawsuit. He is known for being sensitive and holding grudges. After investing in the competing AI company DeepMind but failing to acquire it, he mocked DeepMind founder Demis Hassabis for years. After investing in OpenAI and being rejected when trying to acquire it, he founded the AI company 'X.ai.'

Therefore, Musk's legal issues should be taken with caution. He might want to strike a blow at OpenAI. (OpenAI declined to comment on this.)

Nevertheless, Musk's lawsuit reveals a common phenomenon among the world's top AI companies. These companies start their businesses promising to harness AI's potential for the public good. However, they eventually fall under the influence of big tech corporations.

A case in point is the French AI startup Mistral, dubbed the "European OpenAI." Mistral created AI models as affordably as ChatGPT. Notably, they made them open source so anyone with computing resources could use them for free. The company declared on its homepage a commitment to "fierce independence" and "open and portable" AI. They even once uploaded their products to torrent sites used for downloading illegal content.

However, Mistral recently announced that its latest AI systems are closed source. Only Microsoft Azure cloud customers can access them. MS invested $16 million (about 2.13 billion KRW) in Mistral. While this is not a large portion of the $500 million (about 665.5 billion KRW) Mistral has raised so far, it is enough to attract regulatory attention. Originally, MS's investment in OpenAI was $1 billion (about 1.331 trillion KRW), but it eventually expanded to $13 billion (about 17 trillion KRW), accounting for 49% of the total shares. The core of Musk's lawsuit is that OpenAI surrendered to big tech.

Musk and Altman founded OpenAI in 2015. According to Altman, the goal was to provide AI benefits transparently and openly to humanity and to make the world richer every year. However, over the years, OpenAI became more secretive, and its corporate structure grew complex. In this process, MS became a beneficiary of OpenAI. MS executives announced last year that AI products were on track to become "the fastest-growing $10 billion business in history."

Google's AI division DeepMind took a similar path. Founded 14 years ago to create powerful AI to cure cancer and combat climate change, DeepMind boldly stated on its website for years that it developed AI "to advance science and for the benefit of humanity." However, after being acquired by Google in 2015, and amid the recent generative AI competition, DeepMind's homepage was changed to emphasize products like Gemini.

The goal of curing diseases and enriching everyone seems paradoxically overshadowed by enhancing big tech products that could cause many people to lose their jobs.

Startups like Mistral are being pulled in this direction. Building more powerful AI systems requires massive computing power, which only big tech can reliably access. A Mistral spokesperson said their partnerships "will not lead to compromises on openness."

One might argue this is a rite of passage for startups. The ideals of these startups, which began to make the world a better place, weaken after going public or being acquired. When AI systems are intricately woven into every aspect of life, the risks grow. For example, UK parliament members use generative AI to summarize public feedback. Tools widely used for important decisions should not be controlled by a few opaque oligopolistic companies. It would be disappointing if even Mistral eventually transitions into such a state.

If Musk's lawsuit is effective, it should demand AI companies be honest about their development directions. Musk's lawsuit points out that Altman created a nonprofit board that could fire him if he no longer tried to benefit humanity at OpenAI. After returning, Altman ousted some board members who had fired him. While this is good for OpenAI's business, it would be a betrayal of its benevolent goals. Like the fable "The Emperor's New Clothes," the emperor was not wearing any clothes. And Musk, who said he was "naked," was right.

Fami Olson, Bloomberg Columnist

This article is a translation by Asia Economy of Bloomberg's column "Elon Musk Is Right About OpenAI’s Hypocrisy."