Jan Leike
In May 2024, Jan Leike resigned from OpenAI, where he had led the company's work on AI alignment. Leike had built his career at some of the most prominent institutions in AI safety research. His reasons for leaving were stated plainly in a public letter. What led him to that moment starts at a university in Germany.
Jan Leike earned his undergraduate degree at the University of Freiburg. He completed a master's degree in computer science, then traveled to the Australian National University to pursue a doctorate. His PhD supervisor there was Marcus Hutter. After earning his degree, Leike spent six months as a postdoctoral fellow at the Future of Humanity Institute. From there, he joined DeepMind.
In June 2023, OpenAI launched a project called superalignment, with Jan Leike and Ilya Sutskever as co-leaders. Its stated goal was to determine how to align future artificial superintelligences within four years. The plan was to use relatively advanced AI systems to automate the work of AI alignment research. Sutskever held the title of Chief Scientist at the time, and Leike was Head of Alignment. Time magazine placed Leike on its list of the hundred most influential personalities in AI. He received that recognition in both 2023 and 2024. The four-year timeline the project set for itself would not outlast his tenure at the company.
When Ilya Sutskever left OpenAI, Jan Leike was among those who followed. Daniel Kokotajlo and several other AI safety employees had also departed. Leike put his reasons in writing, stating that 'safety culture and processes have taken a backseat to shiny products.' He added that he had 'gradually lost trust' in OpenAI's leadership. Shortly after leaving, he joined Anthropic, an AI company founded by former OpenAI employees.
Up next
Continue browsing
Common questions
Why did Jan Leike resign from OpenAI?
Jan Leike wrote in his resignation statement that 'safety culture and processes have taken a backseat to shiny products.' He also stated that he had 'gradually lost trust' in the company's leadership. His departure in May 2024 followed the exits of Ilya Sutskever, Daniel Kokotajlo, and several other AI safety employees.
What was the OpenAI superalignment project that Jan Leike co-led?
The superalignment project was launched by OpenAI in June 2023, with Jan Leike and Ilya Sutskever as co-leaders. Its goal was to determine how to align future artificial superintelligences within a four-year window, using relatively advanced AI systems to automate the research process.
Where did Jan Leike go after leaving OpenAI?
Jan Leike joined Anthropic in May 2024, shortly after announcing his resignation from OpenAI. Anthropic is an AI company founded by former OpenAI employees.
Where did Jan Leike study and who supervised his PhD?
Jan Leike completed his PhD in machine learning at the Australian National University under the supervision of Marcus Hutter. He had previously earned an undergraduate degree at the University of Freiburg in Germany and a master's degree in computer science.
What role did Jan Leike hold at OpenAI?
Jan Leike was the Head of Alignment at OpenAI. He co-led the superalignment project alongside Ilya Sutskever, who held the title of Chief Scientist.
What is Jan Leike's background in AI safety research?
Jan Leike is an AI alignment researcher who has worked at DeepMind, OpenAI, and Anthropic. Before joining DeepMind, he spent six months as a postdoctoral fellow at the Future of Humanity Institute, and at DeepMind he conducted empirical AI safety research in collaboration with Shane Legg.
All sources
8 references cited across the entry
- 2MagazineTIME100 AI 2023: Jan Leike7 September 2023
- 3Introducing SuperalignmentJan Leike et al. — 5 July 2023
- 4MagazineTIME100 AI 2024: Jan LeikeHarry Booth — 2024-09-05
- 5"I lost trust": Why the OpenAI team in charge of safeguarding humanity implodedSigal Samuel — 17 May 2024
- 6OpenAI's AI safety teams lost at least seven researchers in recent monthsMatthias Bastian — 18 May 2024
- 7NewsOpenAI putting 'shiny products' above safety, says departing researcherDan Milmo — 18 May 2024