Section 3: Obligations of Providers of General-Purpose AI Models with Systemic Risk

What this part of the EU AI Act covers

Section 3 adds the obligations that apply only to general-purpose AI models with systemic risk: model evaluation including adversarial testing, assessment and mitigation of systemic risks at Union level, tracking and reporting of serious incidents, and an adequate level of cybersecurity protection.

It binds only providers of models classified as carrying systemic risk, and the duties are ongoing rather than pre-market: continuous evaluation, adversarial testing, risk mitigation at Union level, incident tracking and reporting, and cybersecurity protection.