The EU's AI Act establishes a risk-based hierarchy for AI use cases. It bans certain harmful practices, requires conformity assessments for high-risk uses, and sets transparency obligations for medium-risk AI systems. The Act also introduces specific requirements for General Purpose AI models.