Ilya Sutskever, one of the co-founders of OpenAI, has founded a new company, Safe Superintelligence Inc. (SSI), just one month after officially leaving OpenAI.
Sutskever, the longtime chief scientist at OpenAI, founded SSI with former Y Combinator partner Daniel Gross and former OpenAI engineer Daniel Levy.
At OpenAI, Sutskever was integral to the company's efforts to improve AI safety with the rise of “super-intelligent” AI systems, an area in which he worked alongside Jan Rijke, who co-led OpenAI's Superalignment team. But after clashing with OpenAI's leadership over how to approach AI safety, both Sutskever and Rijke dramatically left the company in May; Rijke now leads a team at rival AI shop Anthropic.
Sutskever has long been highlighting the troubling aspects of AI safety: In a blog post published in 2023, he and Reicke predicted that superhumanly intelligent AI would emerge within the decade, and that even if it does come to fruition, it won't necessarily be benevolent, necessitating research into how to control and limit it.
Superintelligence is within reach.
Building secure superintelligence (SSI) is the most important technological challenge of our time.
We launched the world's first Straight Shot SSI Lab with one goal and one product: safe superintelligence.
It's called “Safe Superintelligence”…
— SSI Inc. (@ssi) June 19, 2024
He remains as passionate about the cause today as he ever was. A Wednesday afternoon tweet announcing the formation of Sutskever's new company read, “SSI is our mission, our name, our entire product roadmap — because it's our sole focus. Our team, investors, and business model are all aligned to make SSI a reality. We are tackling safety and functionality in parallel as technical problems to be solved through innovative engineering and scientific breakthroughs.”
“We plan to evolve our capabilities as quickly as possible while always keeping safety top of mind. This way we can scale with peace of mind. Because we're focused, we aren't distracted by administrative costs or product cycles, and our business model ensures that safety, security and advancements are all insulated from short-term commercial pressures.”
Sutskever spoke to Bloomberg in more detail about the new company, but did not discuss its funding or valuation.
What's even clearer is that unlike OpenAI, which launched as a nonprofit in 2015 and then restructured when it became clear its computing power required huge amounts of funding, SSI is being designed from the ground up as a for-profit. Judging by the interest in AI and the team's track record, SSI, too, may soon find itself in financial trouble. “Of all the problems we face, fundraising is not going to be one of them,” Gross told Bloomberg.
SSI has offices in Palo Alto and Tel Aviv and is currently hiring technical talent.