Understand why general-purpose models have model-level obligations.
General-purpose AI models are capable of performing a wide range of tasks and are regulated at model level in addition to the rules that may apply to downstream AI systems.
Providers of GPAI models have baseline documentation and information duties. Additional obligations apply to GPAI models with systemic risk.
A key quantitative presumption for systemic risk is associated with very high training compute, while designation can also consider broader capability and impact factors.
A client uses a third-party foundation model through an API. List the questions needed to determine the client's role and its dependence on the upstream provider.
Based on the uploaded EU AI Specialist lesson notes and checked against the consolidated EU AI Act in force on 27 July 2026. Where the source pack and current law differ, the current law wins.