OpenAI whistleblower found dead at 26 in San Francisco apartment

According to the San Francisco Office for the Chief Medical Examiner, a former OpenAI employee named Suchir Balaji was found dead in his San Francisco home. The New York Times interviewed the 26-year old AI researcher in October. He expressed concerns about OpenAI’s violation of copyright laws.

The Office of the Chief Medical Examiner has identified the deceased as Suchir Balaji. He is 26 years old and lives in San Francisco. A spokesperson told TechCrunch that the cause of death was suicide. The OCME has notified next-of-kin, and at this time, it has no further comments or reports to publish.

Balaji, who worked at OpenAI for nearly four years, quit the company after realizing that the technology would do more harm than good in society, he told The New York Times. Balaji was most concerned about the way OpenAI used copyright data and believed that its practices were harmful to the internet. OpenAI’s spokesperson told TechCrunch in an email that they were “devastated” to hear the news. Their hearts went out to Suchir and his family during this difficult time.

Balaji’s body was discovered in his Buchanan Street flat on November 26, according to a spokesperson from the San Francisco Police Department. Officers and medics were summoned to his Lower Haight apartment to perform a health check on the former OpenAI research. According to police, no evidence of foul play has been found during the initial investigation.

Balaji said in a tweet dated October that he had been at OpenAI for almost 4 years, and had worked on ChatGPT during the last 1.5. “I didn’t know a lot about copyrights, fair use, etc. at first. After seeing the lawsuits against GenAI companies, I became curious. I was able to better understand the issue and came to the conclusion fair use is a pretty implausible defence for many generative AI products. This is because they can create substitutes which compete with the data that they are trained on.

Balaji’s death was reported first by the San Jose Mercury News. OpenAI and Microsoft have been involved in several ongoing lawsuits by newspapers and media publishers including the New York Times who claim that the generative AI startup violated copyright laws.

A court filing on November 25, a day before police discovered Balaji’s corpse, named the former OpenAI worker in a copyright suit brought against the startup. OpenAI agreed, as part of a compromise in good faith, to search Balaji’s file relating to the copyright concerns that he had raised recently.

While several former OpenAI employees raised concerns about the safety culture of the startup, Balaji was the only one who questioned the data used to train OpenAI’s models. In a blog entry from October, the OpenAI researcher said that he did not believe ChatGPT made a fair use its training data. However, he added, similar arguments could also be made for other generative AI products.

The 26-year old researcher studied computer science at University of California, Berkeley before working at OpenAI. He interned at OpenAI during college and Scale AI. The former is where he will work.

Balaji was working on WebGPT in his early days with OpenAI. It was a fine-tuned GPT-3 version that could search the internet. OpenAI released an early version SearchGPT earlier this year. Balaji later worked with the reasoning team at o1, on the GPT-4 pre-training team, and on ChatGPT post-training, according to LinkedIn.

Balaji’s former colleagues and peers in the AI field took to social media sites to mourn his death.

View Bio