OpenAI’s former chief scientist has a new startup

When former OpenAI chief scientist Ilya Sutskever left the company in May, everyone wondered why.

In fact, the recent internal turmoil at OpenAI and a short-lived lawsuit by early OpenAI backer Elon Musk were suspicious enough for the internet hivemind to come up with the “What did Ilya see” meme, referring to the theory that Sutskever saw something alarming in the way CEO Sam Altman led OpenAI.

Now, Sutskever has a new company, and it may be a hint at why, exactly, he left OpenAI at the perceived height of its power. On Wednesday, Sutskever tweeted that he’s starting a company called Safe Superintelligence.

“We will pursue safe superintelligence in a straight shot, with one focus, one goal, and one product. We will do it through revolutionary breakthroughs produced by a small cracked team,” wrote Sutskever.

Tweet may have been deleted

The company’s website is currently just a text message signed by Sutskever as well as co-founders Daniel Gross and Daniel Levy (Gross was a co-founder of search engine Cue, which was acquired by Apple in 2013, while Levy ran the Optimization team at OpenAI). The message reiterates safety as the key component of building an artificial superintelligence.

“We approach safety and capabilities in tandem, as technical problems to be solved through revolutionary engineering and scientific breakthroughs. We plan to advance capabilities as fast as possible while making sure our safety always remains ahead,” the message reads. “Our singular focus means no distraction by management overhead or product cycles, and our business model means safety, security, and progress are all insulated from short-term commercial pressures.”

While Sutskever never publicly explained why he left OpenAI, instead praising the company’s “miraculous” trajectory, it’s notable that safety is at the centre of his new AI product. Musk and several others warned that OpenAI is reckless about building AGI (artificial general intelligence), and the very departure of Sutskever and others in OpenAI’s safety-focused team indicate the company may have been lax when it comes to making sure AGI is being built in a safe way. Musk also has beef with Microsoft’s involvment in OpenAI, claiming that the company has been transformed from an nonprofit into a “closed-source de facto subsidiary” of Microsoft.

In an interview with Bloomberg, published on Wednesday, Sutskever and co-founders did not name any backers, though Gross said that raising capital is not going to be a problem for the startup. It’s also unclear whether SSI’s work will be published as open source.