Frontier AI Model

A regulatory or policy category for the most capable AI models available at a given time, which are subject to extra reporting, evaluation or safety obligations because of the scale of risk they could pose if misused or misaligned.