Technical AI Safety
Research talks and panels on interpretability, evaluations, scalable oversight, and deceptive model behavior, including an evaluations panel with Apollo Research, Arcadia Impact, Palisade Research, and UK AISI.
The 2025 Edition
The first Zurich AI Safety Day brought together researchers, engineers, policymakers, and newcomers from across Europe for a full day of talks, workshops, and high-density networking.
Impact
Outcomes at a Glance
210
Attendees
21
AI Safety Organizations
35
Organization Representatives
4.6/5
Average Rating
330+
Meaningful Connections Made
74%
Rated It a Far Better Use of Time Than Alternatives
Swiss AI Weeks 2025
Best Event
Recognized among more than 230 events across Switzerland
Sessions
The Program
The day featured keynotes, panels, workshops, and lightning talks from speakers at organizations including Anthropic, Apollo Research, UK AISI, FAR.AI, the Future of Life Institute, the ETH AI Center, ERA, Successif, HIP, and Palisade Research, organized around three tracks. A two-hour organization fair and structured networking rounded out the program.
AI Governance
Sessions on the policy landscape, from EU policymaking experience to technical roadmaps for effective AI governance.
Field Building & Careers
Career-transition guidance, personalized advising roundtables, and pathways into AI safety for professionals and newcomers alike.
Archive
Read More
- Learnings from the Zurich AI Safety Day Our detailed write-up of what worked, what didn't, and what we measured.
- Event Wrap-Up on LinkedIn Impressions and highlights shared after the event.
Next edition
Join Us in 2026
The second edition takes place on 7–8 November 2026 at ETH Zurich, expanding to two full days.