Paper “Optimizing for Persuasion Improves LLM Generalization” accepted at NeurIPS 2025 Workshop on Multi-Turn Interactions in Large Language Models!