Peter Eckersley
Peter Eckersley was a computer scientist whose career connected internet security work with AI ethics and safety research. At EFF, he worked on privacy tools such as Panopticlick, HTTPS Everywhere, Certbot, Privacy Badger, and the SSL Observatory, and he co-founded Let’s Encrypt. He completed a PhD in computer science and law, co-authored a report on malicious AI use, led the Partnership on AI’s research, and worked on SafeLife and AI value alignment. He co-founded the AI Objectives Institute before his death in 2022, and the Internet Hall of Fame named him a posthumous 2023 inductee.
Timeline
-
2006
A Melbourne degree and EFF start
After graduating from the University of Melbourne, where he majors in mathematics, physics, and computer science, he joins EFF as a staff technologist in 2006. He later becomes EFF’s Chief Computer Scientist and works on privacy and security projects including Panopticlick, HTTPS Everywhere, Certbot, Privacy Badger, and the SSL Observatory.
-
2012–2015
A PhD and Let’s Encrypt
He completes a PhD in computer science and law at the University of Melbourne in 2012. At the RSA Conference in 2012, he and Alex Halderman plan an automated system for HTTPS certificates. He co-founds Let’s Encrypt, which begins service in 2015.
-
2017–2019
AI value alignment and malicious use
He presents Impossibility and Uncertainty Theorems in AI Value Alignment as a NIPS poster in 2017. In 2018, he coauthors The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation, a report that surveys AI security threats and proposes ways to forecast, prevent, and mitigate them. At SafeAI in 2019, he presents the value-alignment paper, which proposes conjectures about the relationship between uncertainty in objectives and severe unintended consequences.
-
2018–2020
Director of Research at PAI
In September 2018, he leaves EFF to become the Partnership on AI’s inaugural Director of Research, a role he holds until 2020. He leads research on AI ethics, safety, fairness, inclusiveness, trust, and robustness.
-
2019
SafeLife for reinforcement learning
With Carroll Wainwright, he releases SafeLife in 2019, a reinforcement-learning environment that tests how agents avoid unwanted side effects while completing tasks.
-
2021
Co-founding AI Objectives Institute
In 2021, he co-founds the AI Objectives Institute, a non-profit research lab. The institute works to ensure that AI and future economic systems are built and deployed with human objectives at their core.
-
2022
Death after cancer treatment
He dies in September 2022 from complications of cancer treatment.
-
2023
Internet Hall of Fame induction
The Internet Hall of Fame names him a posthumous 2023 inductee for helping expand HTTPS adoption.
Career
- From 2021AI Objectives InstituteCo-founder
- 2018–2020Partnership on AIDirector of Research
- From 2012Let's EncryptCo-founder
- 2006–2018Electronic Frontier FoundationStaff Technologist; Chief Computer Scientist
Education
- 2012University of MelbournePhD
Colleagues
Sources and credits
Photo: We have no published photo of Peter Eckersley yet. Send us a free-licensed photo.
- Peter Eckersley - Internet Hall of Fame
- Peter Eckersley (1978-2022) | Electronic Frontier Foundation
- Partnership on AI Announces Peter Eckersley as Director of Research - Partnership on AI
- Remembering Peter Eckersley
- Impossibility and Uncertainty theorems in AI Value Alignment
- Impossibility and Uncertainty Theorems in AI Value Alignment (or why your AGI should not have a utility function)
- The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation
- Introducing SafeLife: Safety Benchmarks for Reinforcement Learning
- AI • Objectives • Institute
- Champions of Internet Security, Accessibility, and Global Connectivity: 2023 Internet Hall of Fame Inductees Announced