Roles are system-specific
Do not label an entire company simply as a provider or deployer. Map the role for each system and activity. A company can hold several roles across its portfolio.
Use this page to avoid relying on older course dates or simplified memory cues. It is a training reference, not legal advice. For live client work, open the authoritative source and confirm the current wording.
Do not label an entire company simply as a provider or deployer. Map the role for each system and activity. A company can hold several roles across its portfolio.
Prohibited practices are a separate Article 5 analysis. High-risk systems can be permitted but face extensive requirements and obligations when the relevant provisions apply.
The same technology can produce different regulatory outcomes depending on what it is used to do, who it affects and the decision context.
A system does not need to be high-risk before a transparency duty can arise. Analyse the specific interaction or content scenario and the relevant provider/deployer duty.
Map model-level GPAI responsibilities separately from the roles and obligations of a downstream AI system that incorporates the model.
The training-compute threshold creates an important quantitative presumption, while the Commission can also designate a GPAI model based on broader capability and impact criteria.
Conformity-assessment routes depend on the system category and applicable product framework. Do not tell a client that all high-risk AI requires third-party approval.
An AIMS can provide repeatable governance, evidence and continual improvement. Certification does not by itself prove compliance with every EU AI Act obligation.
The uploaded EU AI Specialist lesson notes remain a core learning source in the Academy. Where an older lesson statement conflicts with current legislation or current official guidance, the current authoritative source should be used for client work.