The rapid growth rate of AI combined with its unchecked harms creates a risk of negative future outcomes

The field of preventing harm from advanced AI systems, spanning alignment, evaluation, and decisions to slow or withhold models whose capabilities outpace their safeguards.