OpenAI Says It’s “Devastated” That a Whistleblower Against It Has Been Found Dead

"Our hearts go out to Suchir’s loved ones during this difficult time."

DOA

Soon after blowing the whistle on OpenAI's alleged use of copyrighted material in its training data, one of its former researchers was found dead of an apparent suicide.

In a statement to CNBC, a spokesperson for the firm said that OpenAI was "devastated" to hear of the untimely passing of its former employee Suchir Balaji, a 26-year-old who died in his San Francisco apartment in recent weeks.

"We are devastated to learn of this incredibly sad news today," the firm told CNBC, "and our hearts go out to Suchir’s loved ones during this difficult time."

According to the Bay Area's Mercury News — which, along with the New York Times, is suing OpenAI for alleged copyright infringement — Balaji's body was found on November 26 after someone requested a wellness check at his apartment. While it's unclear who made the call or why they put in the request, authorities currently believe there was "no evidence of foul play" and that his death was indeed a suicide.

Still, the youthful whistleblower's body was found almost exactly a month after the NYT published his allegations, which include Balaji's claim that his job was essentially hoovering up copyrighted material to train OpenAI's models without the consent of its creators.

Silicon Valley'd

A native of Cupertino, California, the Silicon Valley town that hosts Apple's headquarters, Balaji fell in love with AI after learning about a Google DeepMind neural network that had mastered the ancient Chinese game "Go."

"I thought that AI was a thing that could be used to solve unsolvable problems, like curing diseases and stopping aging," he told the NYT. "I thought we could invent some kind of scientist that could help solve them."

After matriculating at UC Berkeley, the young man became one of the lucky grads at his alma mater who got the chance to work at OpenAI in 2020. A few years in, he began working to train the still-unreleased GPT-4 large language model (LLM).

Although the material he was feeding into it was copyrighted, he and his colleagues thought of it more as a "research project" despite the company's for-profit status — an assessment he later came to believe wasn't right.

Eventually, Balaji determined that the material he was feeding into GPT-4 was not "fair use" after all and said as much on his personal blog. He left OpenAI this August because, as he said, "If you believe what I believe, you have to just leave the company."

It's impossible to know what was going on in Balaji's mind when he died. Nevertheless, his death casts a pall over the firm, which has in the last year become the subject of ample media drama as its technology — and its client roster — gets more and more powerful.

More on ex-OpenAI-ers: AI Safety Researcher Quits OpenAI, Saying Its Trajectory Alarms Her