The finalized framework requires signatory nations and companies to disclose compute usage above a defined threshold for frontier AI model training.
How Enforcement Will Work
Under the new framework, companies training frontier models above the disclosure threshold must file detailed compute usage reports with an independent oversight body, with penalties for non-compliance including public disclosure of violations.
Several major AI labs have already begun voluntary early reporting ahead of the framework's formal rollout.
#AISummit#ComputeDisclosure
