An AI Safety Institute (AISI), in general, is a state-backed institute aiming to evaluate and ensure the safety of the most advanced artificial intelligence (AI) models, also called frontier AI models.