Definition
The government process for deciding which advanced (frontier) AI models get scrutinised before release and who is trusted to conduct that review. When this process is withheld or opaque, the companies it governs cannot plan around it.
Why it matters
If a company cannot predict whether its model will face binding pre-release review, or who will run it, it cannot plan launches, compliance or liability — and this uncertainty is now spilling into proposed binding testing mandates.