Stay up to date with free updates
Just sign up artificial intelligence myFT Digest – delivered straight to your inbox.
Just a month after leaving the company following the attempted coup against CEO Sam Altman, OpenAI co-founder Ilya Sutskever is launching a rival AI startup focused on “building safe superintelligence.”
Sutskever, one of the world's most respected AI researchers, on Wednesday founded Safe Superintelligence (SSI), which describes itself as “the world's first full-scale SSI lab with one goal and one product: safe superintelligence,” according to a statement posted on X.
Sutskever co-founded the independent US startup with former OpenAI employee Daniel Levy and AI investor and entrepreneur Daniel Gross, who was a partner at Y Combinator, the Silicon Valley startup incubator that Altman previously ran.
Gross has invested in GitHub, Instacart, and other AI companies including Perplexity.ai, Character.ai and CoreWeave. SSI's investors were not disclosed.
They said developing safe superintelligence — a type of machine intelligence that can surpass human cognitive capabilities — is the company's “sole focus,” and added that the company is not bound by revenue demands from investors, including by bringing in top talent to work on the effort.
OpenAI had a similar mission to SSI when it was founded in 2015 as a nonprofit research institute to develop superintelligent AI for the benefit of humanity, and Altman maintains that this is still OpenAI's guiding principle, but the company has grown rapidly under his leadership.
SSI's “single-point focus means we are not distracted by administrative costs or product cycles, and our business model ensures safety, security and progress are all insulated from short-term commercial pressures,” Sutskever and his co-founders said in a statement, adding that the company will be headquartered in both Palo Alto and Tel Aviv.
Sutskever is considered one of the world's leading AI researchers and played a key role in OpenAI's early leadership in the nascent field of generative AI, developing software that can generate multimedia responses to human questions.
The OpenAI split announcement comes after continued turmoil within the leading AI group, centered around disputes over leadership direction and safety.
In November, OpenAI's board of directors, which then included Sutskever, abruptly ousted Altman as CEO, shocking investors and employees. Altman returned a few days later with a new board, but without Sutskever.
After the failed coup attempt, Sutskever stayed on at the company for a few months, but left in May. “I'm excited to see what comes next. This is a project that is very personally meaningful to me, and I will share more details in due course,” she said in her resignation statement.
This isn't the first time OpenAI employees have left the ChatGPT maker to create a “secure” AI system: In 2021, the company's former AI safety chief, Dario Amodei, spun off his own startup, Anthropic, which has raised $4 billion from Amazon and hundreds of millions more from venture capitalists, giving it a valuation of more than $18 billion, according to people familiar with the talks.
While Sutskever has professed confidence in OpenAI's current management, Jan Reicke, who recently left the company and worked closely with Sutskever, said his differences with the company's management had “reached a breaking point” and that “safety culture and processes were taking a backseat to flashy products.” Reicke has since joined Anthropic, an OpenAI rival.
