Fellowships
Our first fellowship runs this autumn, on where rapid AI progress is heading, and what it would take to handle it well.
AaltoAI / Safety works on a hard, unsolved problem in AI: ensuring AI models reliably do what we intend, and knowing when they don't. Frontier AI systems are becoming more autonomous and starting to make high-stakes decisions that affect the real world, while our ability to specify and verify what they're optimising for hasn't kept pace.
We work on exactly that: taking on projects, learning about AI safety through fellowships, hosting research talks, and discussing frontier research in a reading group.
Join us to find out how you can contribute to aligning AI.
Our first fellowship runs this autumn, on where rapid AI progress is heading, and what it would take to handle it well.
Members work together on AI safety projects, from multi-week efforts to one-evening Research Sprints.
Experts in the field, from Finland and abroad, present their research. Talks are open to anyone, no background in safety required.
Since November 2024, we've been hosting a reading group on AI safety and LLMs, where we read and discuss scientific papers and articles. Topics run from reverse-engineering what actually happens inside LLMs to automated scientific discovery.
The AaltoAI / Safety Fellowship is a six-week, in-person programme open to anyone interested in helping ensure that AI development benefits humanity. Together, we will explore how advanced AI might develop, the risks it could pose, and what it could take to make its development go well.
During the first five weeks, participants work through weekly readings and take part in group discussions, with an expected commitment of around five hours per week. The final week is dedicated to a project that puts the ideas into practice.
The fellowship is based on BlueDot Impact's AGI Strategy curriculum, covering possible paths to advanced AI, approaches to reducing risk, and ways one can contribute.
Anyone interested in helping ensure that AI development benefits humanity, regardless of their field of study or academic background. The fellowship is primarily aimed at Aalto students, but is open to people outside Aalto University too. The final project is tailored to each participant’s skills and experience.
Mon, Sep 28, 2026 at 23:59
More infoWed, Sep 23, 2026 at 17:00
The first five weeks combine independent weekly readings with in-person group discussions. In the sixth and final week, participants complete a project.
Plan for approximately five hours per week during the first five weeks. The time needed for the final-week project will depend on its scope.
The fellowship runs from 5 October to 15 November 2026 and takes place in person on the Aalto University campus.
Participants meet in groups of 5–7 people, at times set by each group’s shared availability.
The curriculum examines how advanced AI might develop, the risks it could pose, approaches to reducing those risks, and ways to help AI development go well.
All curriculum materials are in English. Group discussions will be held in English or Finnish, depending on the shared language of the group. Participants should be comfortable reading and discussing advanced material in English.
No. The fellowship is completely free.
Autumn 2026
Applications are closed
The first cohort is under way.
To hear when the next round opens, follow our announcements channel.
Questions? safety@aaltoai.com