J Rosser

PhD student, University of Oxford

1 paper at NeurIPS 2025

Homepage· OpenReview· Semantic Scholar· Google Scholar

Poster Session 4

1 paper

Thursday, December 4, 2025 · 4:30 PM → 7:30 PM

Exhibit Hall C,D,E

AgentBreeder: Mitigating the AI Safety Risks of Multi-Agent Scaffolds via Self-Improvement

#1201 Spotlight · J Rosser, Jakob Nicolaus Foerster

Our paper introduces AgentBreeder, a framework for multi-objective self-improving evolutionary search over scaffolds, specifically targeting scaffolds' safety impact on large language models in multi-agent systems.