Ilya Sutskever

Toronto, AlexNet and Google

Ilya Sutskever was a student in the machine learning group at the University of Toronto, working with Geoffrey Hinton^[1]. He earned his PhD in computer science from the University of Toronto^[2]. He was also a postdoctoral researcher in Andrew Ng's group at Stanford^[1].

With Alex Krizhevsky and Hinton he wrote ImageNet Classification with Deep Convolutional Neural Networks, presented at NIPS 2012^[3]. The paper describes a deep convolutional network with 60 million parameters trained to classify 1.3 million ImageNet images into 1,000 classes^[3]. The Royal Society describes him as a co-inventor of AlexNet^[2].

Krizhevsky, Sutskever and Hinton incorporated a company, DNNresearch, in 2012^[4]. In March 2013 the University of Toronto announced that Google had acquired it, and that Krizhevsky and Sutskever would move to Google^[4]. Sutskever then spent three years as a research scientist on the Google Brain team^[1].

He co-wrote Sequence to Sequence Learning with Neural Networks with Oriol Vinyals and Quoc Le, posted in September 2014^[5]. NeurIPS gave the paper its Test of Time award in 2024^[6].

OpenAI

When OpenAI was announced in December 2015 as a non-profit research company, Sutskever was its research director^[7]. He later held the title of co-founder and chief scientist^[1][2]. The Royal Society, which elected him a Fellow in 2022, describes him as having led research at OpenAI and as one of the architects behind the GPT models^[2].

TechCrunch reported that at OpenAI he worked on AI safety for superintelligent systems with Jan Leike, who co-led the Superalignment team, and that both left the company in May 2024^[8]. In a 2023 blog post written with Leike, he predicted that AI with intelligence superior to humans could arrive within the decade^[8].

Safe Superintelligence

On 19 June 2024 Sutskever launched Safe Superintelligence Inc. (SSI) with Daniel Gross and Daniel Levy, one month after formally leaving OpenAI^[8]. SSI describes itself as a straight-shot lab with one goal and one product, a safe superintelligence^[9]. It says it approaches safety and capabilities together as technical problems and has offices in Palo Alto and Tel Aviv^[9].

In September 2024 SSI said it had raised $1 billion from NFDG, a16z, Sequoia, DST Global and SV Angel^[10]. On 3 July 2025 Sutskever told staff and investors that Gross had left as of 29 June, that he was now formally CEO and that Levy was president^[10][11]. In the same message he said SSI was focused on its own work despite interest from companies looking to acquire it^[10][11]. In July 2026 SSI announced a strategic partnership with NVIDIA that it said would let it scale its compute by 10 times^[10].

Views on superintelligence and research

At NeurIPS in December 2024, accepting an award, Sutskever said superintelligent systems would be qualitatively different from today's AI^[12]. He predicted they would be agentic in a real way, would reason, and as a result would be more unpredictable^[12].

In a November 2025 interview on the Dwarkesh Podcast, titled We're moving from the age of scaling to the age of research, he discussed SSI's strategy, the problems with pre-training and how to improve generalisation^[13]. In that interview he said current models generalise dramatically worse than people^[13].

Known for

• Co-authoring AlexNet
• Sequence to sequence learning
• Co-founding OpenAI and serving as its chief scientist
• Co-founding and leading Safe Superintelligence

Sources

1. Ilya Sutskever's home page, University of Toronto — checked 6 Nov 2026
2. The Royal Society: Dr Ilya Sutskever FRS — checked 6 Nov 2026
3. NeurIPS Proceedings: ImageNet Classification with Deep Convolutional Neural Networks, 2012 — checked 6 Nov 2026
4. University of Toronto News: Google acquires U of T neural networks company, March 2013 — checked 6 Nov 2026
5. arXiv: Sequence to Sequence Learning with Neural Networks, September 2014 — checked 6 Nov 2026
6. NeurIPS blog: Announcing the NeurIPS 2024 Test of Time paper awards, November 2024 — checked 6 Nov 2026
7. TechCrunch: Nonprofit OpenAI launches, December 2015 — checked 6 Nov 2026
8. TechCrunch: Ilya Sutskever, OpenAI's former chief scientist, launches new AI company, June 2024 — checked 6 Nov 2026
9. Safe Superintelligence Inc.: home page — checked 6 Nov 2026
10. Safe Superintelligence Inc.: Updates — checked 6 Nov 2026
11. TechCrunch: Ilya Sutskever will lead Safe Superintelligence following his CEO's exit, July 2025 — checked 6 Nov 2026
12. TechCrunch: Ilya Sutskever believes superintelligent AI will be unpredictable, December 2024 — checked 6 Nov 2026
13. Dwarkesh Podcast: Ilya Sutskever, We're moving from the age of scaling to the age of research, November 2025 — checked 6 Nov 2026