MSc CS @ ETH Zürich · MATS 10.0 · AI Alignment

Peter Nutter

MSc Computer Science student at ETH Zürich and MATS 10.0 scholar. I work on empirical AI alignment — midtraining interventions, generalization, RL, and reward hacking.

 CV  Google Scholar  GitHub  Email

Scroll

Recent updates

MATS 10.0 Jun–Aug 2026 · Berkeley

Incoming Scholar, Technical Track. Working on extending Model Spec midtraining.

ERA Fellowship Feb–May 2026 · Cambridge

Independent empirical AI safety research on reward hacking in code models — when it emerges, whether it generalizes, and whether it can be detected.

ICML 2026 · Spotlight Position paper · main author

“Anthropomorphic Misalignment Research Needs Stronger Evidence” (spotlight + oral).

Selected research

Notes & courses