También: alignment, ai alignment

Work on making AI systems behave as intended. Covers alignment research, red teaming and evaluations, misuse prevention, interpretability, and deployment policy for capable models.