Article 52: GPAI Models with Systemic Risk (>10^25 FLOPs)
Article 52 classifies frontier foundation models with high-impact capabilities (cumulative training compute exceeding 10^25 FLOPs) as carrying Systemic Risk, subject to stringent red-teaming.
Binding European Union legislative enactment across all 27 Member States.
Calculated on consolidated global group turnover under Article 71 formulas.
Requires specialized corporate tech E&O riders for regulatory defense.
Codified Statutory Requirements & Clauses
Detailed clause-by-clause legal and engineering specifications under Article 52.
Presumption of systemic risk if cumulative training compute exceeds 10^25 FLOPs or determined by the EU AI Office based on capability evaluations.
Continuous model evaluations including conducting and documenting adversarial testing (red-teaming) to identify and mitigate systemic risks.
Immediate reporting of serious cybersecurity incidents, catastrophic failures, or critical vulnerabilities to the EU AI Office.
Article 52 Technical Compliance Protocol
Complete this operational verification protocol to ensure engineering and organizational alignment with Article 52.