AI Safety Göttingen

AI Safety Göttingen

We are a technical reading group at University of Göttingen, gathering biweekly to discuss papers and blogs posts in AI safety.

Topics of interest include: emergent misalignment, benchmarks, alignment metrics, multi-agent interactions, chain of thought monitoring, scaling and autonomy (recursive self improvement, continual learning, long-horizon tasks), LLMs as moral patients, scary new tech, famous incidents, and many others (open to suggestions).

Join the WhatsApp Group See upcoming meetings
The Gänseliesel statue of Göttingen facing a mechanical goose

Upcoming Meetings

  • Mon 26 Oct

    AI Safety Evals and Benchmarks

    Monday, October 26 · 17:30

  • Mon 9 Nov

    The HuggingFace Incident

    Monday, November 9 · 17:30

Location in north campus seminar room, announced in WhatsApp group.

Past Meetings

Resources

METR Substack Nathan Lambert — Interconnects Apollo Research Redwood Research LessWrong Dwarkesh Podcast

Organizers

Sara Rajaram

Sara Rajaram

Sara is an ML PhD student focusing on reinforcement learning and digital twins.

Goirik Chakrabarty

Goirik Chakrabarty

Goirik is an ML PhD student focusing on multimodal AI, vision-to-time series modeling, and robust model evaluation.