Jan Leike (born 1986 or 1987) is an AI alignment researcher with a background in machine learning and reinforcement learning theory. He obtained his undergraduate degree from the University of Freiburg in Germany, followed by a master's degree in computer science. Leike then pursued a PhD in reinforcement learning theory/machine learning at the Australian National University under the supervision of Marcus Hutter. His PhD thesis, "Nonparametric General Reinforcement Learning," was completed in 2016.
After a six-month postdoctoral fellowship at the Future of Humanity Institute, Leike joined DeepMind to focus on empirical AI safety research, where he collaborated with Shane Legg. At DeepMind, he worked as an alignment researcher, prototyping reinforcement learning from human feedback.
In 2021, Leike joined OpenAI, where he co-led the Superalignment Team with Ilya Sutskever, aiming to determine how to align future artificial superintelligences to ensure their safety. This project involved automating AI alignment research using advanced AI systems. During his time at OpenAI, Leike was involved in the development of InstructGPT, ChatGPT, and the alignment of GPT-4. He also co-authored the Superalignment Team's research roadmap. In May 2024, Leike announced his resignation from OpenAI, citing concerns over the company's safety culture and leadership.
Following his departure from OpenAI, Leike joined Anthropic, an AI company founded by former OpenAI employees, in May 2024.
Leike's research aims to solve the "hard problem of alignment," specifically how to train AI systems to follow human intent on tasks that are difficult for humans to evaluate directly. His notable publications include:
Leike has also written a Substack newsletter, "Musings on the Alignment Problem", where he explores various topics related to the alignment problem in AI.
Jan Leike (born 1986 or 1987) is an AI alignment researcher with a background in machine learning and reinforcement learning theory. He obtained his undergraduate degree from the University of Freiburg in Germany, followed by a master's degree in computer science. Leike then pursued a PhD in machine learning at the Australian National University under the supervision of Marcus Hutter, completing his thesis on "Nonparametric General Reinforcement Learning" in 2016.
Leike began his career with a six-month postdoctoral fellowship at the Future of Humanity Institute before joining DeepMind, where he focused on empirical AI safety research and reinforcement learning from human feedback, collaborating with Shane Legg.
In 2021, Leike joined OpenAI, where he played a pivotal role in the development of groundbreaking technologies. In June 2023, he became the co-leader of the "superalignment" project alongside Ilya Sutskever, with the goal of ensuring the safety of future artificial superintelligences. This project involved automating AI alignment research using advanced AI systems. Leike's work at OpenAI also included the development of InstructGPT and ChatGPT, contributing to significant advancements in the field.
In May 2024, Leike announced his resignation from OpenAI, citing concerns about the company's safety culture and leadership. He subsequently joined Anthropic, an AI startup founded by former OpenAI employees, where he continues to pursue his research interests in superalignment.
Leike has an impressive list of publications to his name, including:
Leike's research primarily focuses on the "hard problem of alignment," aiming to train AI systems to follow human intent on complex tasks that are challenging for direct human evaluation. His work has gained widespread recognition, and in 2023, he was listed as one of TIME's 100 most influential people in AI.
Jan Leike is a prominent figure in the field of AI alignment and machine learning, known for his significant contributions to the development of safe and ethical AI technologies. His research and publications have advanced the understanding of AI alignment, and his work continues to shape the future of artificial intelligence.
Youtube Title: OpenAI’s huge push to make superintelligence safe | Jan Leike
Youtube Link: link
Youtube Channel Name: 80,000 Hours
Youtube Channel Link: https://www.youtube.com/@eightythousandhours
OpenAI’s huge push to make superintelligence safe | Jan Leike
Youtube Title: Jan Leike | Super Intelligent Alignment @ Intelligent Cooperation Workshop
Youtube Link: link
Youtube Channel Name: Foresight Institute
Youtube Channel Link: https://www.youtube.com/@ForesightInstitute
Jan Leike | Super Intelligent Alignment @ Intelligent Cooperation Workshop
Youtube Title: Eine deutliche Warnung
Youtube Link: link
Youtube Channel Name: Karl Olsberg
Youtube Channel Link: https://www.youtube.com/@KarlOlsbergAutor
Eine deutliche Warnung
Youtube Title: Ilya Sutskever and Jan Leike RESIGN from OpenAI - My in-depth analysis - end of an era!
Youtube Link: link
Youtube Channel Name: David Shapiro
Youtube Channel Link: https://www.youtube.com/@DaveShap
Ilya Sutskever and Jan Leike RESIGN from OpenAI - My in-depth analysis - end of an era!
Youtube Title: Sam Altman WRECKS OpenAI - Jan Leike joins Anthropic - Brain Drain from OpenAI
Youtube Link: link
Youtube Channel Name: David Shapiro
Youtube Channel Link: https://www.youtube.com/@DaveShap
Sam Altman WRECKS OpenAI - Jan Leike joins Anthropic - Brain Drain from OpenAI
Youtube Title: Jan Leike, OpenAI (MIT AI Event)
Youtube Link: link
Youtube Channel Name: AttentionX
Youtube Channel Link: https://www.youtube.com/@attentionx
Jan Leike, OpenAI (MIT AI Event)
Youtube Title: Googler Reacts To David Shapiro's: Sam Altman WRECKS OpenAI - Jan Leike joins Anthropic
Youtube Link: link
Youtube Channel Name: SVIC Podcast
Youtube Channel Link: https://www.youtube.com/@svicpodcast
Googler Reacts To David Shapiro's: Sam Altman WRECKS OpenAI - Jan Leike joins Anthropic
Youtube Title: OpenAI Faces Turmoil as Jan Leike Resigns: What Does this Mean for the Future of AI?
Youtube Link: link
Youtube Channel Name: AI Insight News
Youtube Channel Link: https://www.youtube.com/@AIInsightNews
OpenAI Faces Turmoil as Jan Leike Resigns: What Does this Mean for the Future of AI?
Youtube Title: Jan Leike - AI alignment at OpenAI
Youtube Link: link
Youtube Channel Name: Towards Data Science
Youtube Channel Link: https://www.youtube.com/@TowardsDataScience
Jan Leike - AI alignment at OpenAI
Youtube Title: is GPT dangerous? The Truth behind Jan Leike's Move from OpenAI to Anthropic
Youtube Link: link
Youtube Channel Name: AI Untold
Youtube Channel Link: https://www.youtube.com/@aiuntold_
is GPT dangerous? The Truth behind Jan Leike's Move from OpenAI to Anthropic
Youtube Title: Stanford CS25: V2 I Language and Human Alignment
Youtube Link: link
Youtube Channel Name: Stanford Online
Youtube Channel Link: https://www.youtube.com/@stanfordonline
Stanford CS25: V2 I Language and Human Alignment
Youtube Title: Jan Leike - Scaling Reinforcement Learning from Human Feedback
Youtube Link: link
Youtube Channel Name: FAR AI
Youtube Channel Link: https://www.youtube.com/@FARAIResearch
Jan Leike - Scaling Reinforcement Learning from Human Feedback
Youtube Title: Jan Leike – General Reinforcement Learning – CSRBAI 2016
Youtube Link: link
Youtube Channel Name: Machine Intelligence Research Institute
Youtube Channel Link: https://www.youtube.com/@MIRIBerkeley
Jan Leike – General Reinforcement Learning – CSRBAI 2016
Youtube Title: 80,000 Hours - Jan Leike on OpenAI’s massive push to make superintelligence safe in 4 years or less
Youtube Link: link
Youtube Channel Name: IMM
Youtube Channel Link: https://www.youtube.com/@imm_radio
80,000 Hours - Jan Leike on OpenAI’s massive push to make superintelligence safe in 4 years or less
Youtube Title: Is OpenAI Neglecting Safety Concerns? (Jan Leike Adresses Issues)
Youtube Link: link
Youtube Channel Name: WieseTechnology
Youtube Channel Link: https://www.youtube.com/@WieseTechnology
Is OpenAI Neglecting Safety Concerns? (Jan Leike Adresses Issues)
Youtube Title: AGI-15 Jan Leike - Using Localization and Factorization to Reduce the Complexity of RL
Youtube Link: link
Youtube Channel Name: AGI Society
Youtube Channel Link: https://www.youtube.com/@AGISocietyOfficial
AGI-15 Jan Leike - Using Localization and Factorization to Reduce the Complexity of RL
Youtube Title: 🚩OpenAI Safety Team "LOSES TRUST" in Sam Altman and gets disbanded. The "Treacherous Turn".
Youtube Link: link
Youtube Channel Name: Wes Roth
Youtube Channel Link: https://www.youtube.com/@WesRoth
🚩OpenAI Safety Team "LOSES TRUST" in Sam Altman and gets disbanded. The "Treacherous Turn".
Youtube Title: Microsoft Promises a 'Whale' for GPT-5, Anthropic Delves Inside a Model’s Mind and Altman Stumbles
Youtube Link: link
Youtube Channel Name: AI Explained
Youtube Channel Link: https://www.youtube.com/@aiexplained-official
Microsoft Promises a 'Whale' for GPT-5, Anthropic Delves Inside a Model’s Mind and Altman Stumbles
Youtube Title: The Exciting, Perilous Journey Toward AGI | Ilya Sutskever | TED
Youtube Link: link
Youtube Channel Name: TED
Youtube Channel Link: https://www.youtube.com/@TED
The Exciting, Perilous Journey Toward AGI | Ilya Sutskever | TED
Youtube Title: Working in AI | Jan Leike, Helen Toner, Malo Bourgon, and Miles Brundage
Youtube Link: link
Youtube Channel Name: Centre for Effective Altruism
Youtube Channel Link: https://www.youtube.com/@EffectiveAltruismVideos
Working in AI | Jan Leike, Helen Toner, Malo Bourgon, and Miles Brundage