The Problem

We know Europe has a rich AI research talent, but the challenge is transforming that talent into AI safety researchers. After reading MIRI's AI safety report in December 2025, I started to think of ways to address this gap.

Fellowships like Safe AI Germany help build the talent pipeline through mentorships and projects, but their pipeline is geared towards people that are already in the field of AI safety. They also do not solve the retention problem because what happens after the fellowship? With Frankfurt AI Safety, I am trying to answer whether we could build the connective tissue around these fellowships, courses, and trainings for the AI Safety talent pipeline in the EU.

What I did

I designed a full annual AI safety curriculum with topics ranging from AI fundamentals to technical safety and governance. I then selected readings for each topic (subject to change based on the most recent and relevant paper per topic) and ran bi-weekly meetups for more than 20 events in 2026, doing speaker sourcing, event facilitation, and marketing around Frankfurt.

I also came up with three projects for the community to run, and we are currently running these projects to bring people hands-on technical AI safety research in compliance with the EU AI Act. Our goal is to publish papers in reputable conferences by the end of the year.

Who is involved

We have speakers, mentors and partners from UNESCO's EU AI Act office, University of Norway, University of Hamburg and Safe AI Germany. We are also looking to involve the industry such as financial institutions to train their staff on AI safety capabilities.

Expected outcome

The outcome is to create a permanent, city-rooted community that generates interest, develops technical talent, and holds the network together long-term. Starting from Frankfurt and scaling across Europe, the goal is to create an environment where AI safety becomes a logical career path for students, researchers, and professionals who would turn their AI safety learnings into real work.