Art. 55—Systemic Risk GPAI Obligations
>Control Description
Providers of general-purpose AI models with systemic risk shall perform model evaluation including adversarial testing, assess and mitigate systemic risks, track and report serious incidents, and ensure adequate cybersecurity protection. Systemic risk models are those with high impact capabilities based on cumulative compute or Commission designation.
Risk Tier
GPAI
Article
Art. 55Deadline
August 2, 2025Penalty
€15M or 3% of global turnover>Applies To
providers
>Key Requirements
- •Model evaluation including adversarial testing
- •Assessment and mitigation of systemic risks
- •Serious incident tracking and reporting
- •Adequate cybersecurity protection
- •Applies to models >10^25 FLOPs training compute
>Cross-Framework Mappings
Ask AI
Configure your API key to use AI features.