PRISM AI Safety Fellowship - Research Projects Spotlights + Q&A

0:01:20 Adversarial Robustness in Protein Foundation Models — Jeyashree Krishnan (Senior ML Scientist, Siemens AG) & Ajay Mandyam Rangarajan (Staff Software Engineer, Synopsys) 0:33:40 Grounding Safe-by-Design AI — Jobst Heitzig (Senior Scientist, Potsdam Institute / Zuse Institute Berlin) 1:01:45 Applying Epidemiological Methods to AI Harm Monitoring — Branwen Owen (Research Fellow, Arcadia Impact) PRISM 2026 is a 16-week remote AI safety research fellowship running June–September 2026. Fellows commit 20 hours/week, working in teams of 4 with a mentor through literature review, methodology proposal, peer voting, experimentation, and paper submission. We have 48 fellow spots across 12 research teams. For the 2025 pilot, 50 fellows produced 4 papers, 2 accepted to NeurIPS and ICASSP. Application closes May 25, 2026 @ 8am PDT. Recorded May 23, 2026 Links: Become a W2D2 Member for FREE: https://bit.ly/join-w2d2 W2D2: http://w2d2.org/ PRISM: http://prism-research.org/