Голова OpenAI Сем Альтман та Сучір Баладжи / YouTube, Instagram
One of the creators of ChatGPT, Suchir Balaji, who exposed issues with OpenAI and artificial intelligence, was found dead in his apartment.
The sudden death of Suchir Balaji shocked the tech world and occurred against the backdrop of heated debates about artificial intelligence and copyright laws. Months before his death, Balaji had been exposing deep ethical and legal problems with AI products. In particular, he criticized how OpenAI and other companies collect massive amounts of data from the internet to train their models. Balaji questioned the legality and ethics of these practices according to current copyright norms.
Suchir Balaji was one of the key architects of ChatGPT. The tragic death of the 26-year-old man, a former member of the OpenAI team and an outspoken critic of the company’s practices, occurred in his San Francisco apartment on November 26. The police confirmed the suicide, with no evidence of foul play at this time.
After four years of working at OpenAI, Suchir Balaji left the company in August 2024, citing concerns about the use of copyright-protected data without proper consent. In October, he discussed this in an interview with The New York Times, to which the engineer approached himself. In the interview, he explained how systems like GPT-4 learn by creating complete copies of data. Balaji detailed his concerns in an essay where he articulated the complexities of AI training processes.
“Although generative models rarely produce output significantly similar to any of their training data, the training process of a generative model involves the creation of copies of copyright-protected content… If these copies are unauthorized, this could potentially be considered a copyright infringement, depending on whether the specific use of the model qualifies as ‘fair use’. Since fair use is determined on a case-by-case basis, it is impossible to make a broad statement about when exactly generative AI meets the requirements of fair use. … The output is not exact replicas of the input data, but they are also not fundamentally new,” explained Balaji.
After leaving, Balaji continued to actively discuss issues related to the ethics and legality of OpenAI’s actions. Just a day before his death, his name “appeared” in a copyright lawsuit filed against the company. Another issue that Balaji debated was the generation of false or completely fabricated information, which he referred to as “hallucinations.” According to him, the internet is changing for the worse.
Balaji shared his concerns in his last post on X Twitter in October and a blog post in the same month. He urged machine learning researchers to learn more about copyright norms.
“At first, I didn’t know much about copyright, fair use, etc., but I became interested when I saw all the lawsuits filed against GenAI companies. As I tried to better understand the issue, I eventually concluded that fair use seems quite unlikely as a defense for many generative AI products for the main reason that they can create substitutes that compete with the data they train on”.
He noted that known precedents, such as the Google Books case, contradict the generative AI companies’ claim of “fair use.” Balaji clarified that his thoughts were directed towards a broader application of generative AI, not just against OpenAI.
Source: TechCrunch