He said he’d “try to testify.”
Spill the Beans
Suchir Balaji, the young OpenAI whistleblower whose death was made public earlier this month, was apparently being considered as a witness against his former employer in a major lawsuit, The Associated Press reports.
Shortly before his passing, the 26-year old Balaji had sounded the alarm on OpenAI’s allegedly illegal copyright practices in an October profile with The New York Times.
But according to the report, his involvement with the newspaper of record wasn’t set to end there. Balaji later told the AP that he would “try to testify” in the strongest copyright infringement cases brought against OpenAI, and considered the NYT’s high-profile one, filed last year, to be the “most serious.”
The Times seems to have had the same idea. In a November 18 court filing, lawyers for the newspaper named Blaji as someone who might possess “unique and relevant documents” that could prove OpenAI knowingly committed copyright infringement.
Tragic Death
Balaji had worked at OpenAI for four years, but quit in August after becoming appalled at what he saw as the ChatGPT developer’s flagrant disregard for copyright law. He had worked first-hand on the company’s massive data scraping efforts, in which it more or less pulled any content it could from the web to train its large language models.
“If you believe what I believe,” Balaji told the NYT, “you have to just leave the company.”
On November 26, a month after his profile in the NYT, Balaji was found dead in his San Francisco apartment, in what the police said was an apparent suicide. His death wasn’t reported until December 13.
Publicly, OpenAI mourned Balaji’s passing. “We are devastated to learn of this incredibly sad news today, and our hearts go out to Suchir’s loved ones during this difficult time,” a company spokesperson told CNBC at the time.
Following Suit
The high-profile lawsuit that Balaji was being considered as a witness for was filed by the NYT last December, alleging that OpenAI had illegally used the newspaper’s copyrighted work to train its chatbots. Balaji’s documents were also being sought by another suit filed by comedian Sarah Silverman against OpenAI and Meta, the AP said.
OpenAI and other tech companies argue that their use of copyrighted data on the internet constitutes “fair use” because their AI models significantly transform that content. But Balaji disagreed, saying that the AI models create a copy of the data they ingest, and are from there instructed to generate text of dubious originality.
“The outputs aren’t exact copies of the inputs, but they are also not fundamentally novel,” he told the NYT in October.
Balaji’s family said that a memorial is being planned for later this month at the India Community Center in Milpitas, California.
More on AI: OpenAI Whistleblower Disgusted That His Job Was to Vacuum Up Copyrighted Data to Train Its Models