EU GPAI Code
Voluntary Code designed to help general-purpose AI model providers demonstrate compliance with AI Act obligations concerning transparency, copyright and systemic-risk management.
Voluntary Code designed to help general-purpose AI model providers demonstrate compliance with AI Act obligations concerning transparency, copyright and systemic-risk management.
Primarily relevant to providers of general-purpose AI models, with additional systemic-risk commitments for the most capable models.
The Code is a voluntary compliance tool; underlying AI Act obligations remain legally binding.
Penalty context: Non-participation is not itself the same as infringement; underlying AI Act duties and enforcement remain applicable.
Voluntary Code designed to help general-purpose AI model providers demonstrate compliance with AI Act obligations concerning transparency, copyright and systemic-risk management.
Maintain an operational policy addressing applicable EU copyright requirements for model development and use.
Actor: providers · Domain: governance
Evidence: Policy, process evidence and escalation records.
Maintain structured model information needed by downstream providers and authorities.
Actor: providers · Domain: documentation
Evidence: Model card, capability/limitation documentation and update history.
For models in scope, identify, assess, mitigate and monitor systemic risks across the model lifecycle.
Actor: providers · Domain: risk management
Evidence: Risk model, evaluations, mitigations, monitoring and incident evidence.
Cybatar's mapping is designed for AI governance and assurance work. Legal interpretation remains anchored to the current primary text and competent authority guidance.
Open European Commission source