Aligning AI Teams New research examines why alignment training applied to individual LLM agents fails to transfer to multi-agent settings and tests interventions aimed at preserving that transfer in multi-agent LLM teams. The work focuses on two software-engineering tasks and investigates the underlying mechanisms behind the failure. Previous work has shown that alignment training on individual LLM agents can fail to transfer to multi-agent settings. We investigate the underlying mechanisms for this failure, and explore interventions to preserve alignment transfer to multi-agent LLM teams. We focus on two software-engineering ta