How to classify an AI system under the AI Act: the prohibited practices, the two high-risk routes, the Article 6(3) exemption, transparency-only systems and minimal risk, and general-purpose AI models, with the date each tier applies after the 2026 Digital Omnibus.
What a high-risk AI system needs before and after it reaches the EU market: the Articles 8 to 15 requirements, provider and deployer duties, retention periods, incident deadlines, the 2027 and 2028 application dates, legacy systems, standards and fines.
Every AI Act application date as of October 2026, including the changes made by the Digital Omnibus on AI (Regulation (EU) 2026/1744): what already applies, what starts in December 2026, the new 2027 and 2028 high-risk dates, and the transition periods.
How the AI Act splits duties between providers and deployers, when a deployer, importer or distributor becomes a provider, what each role owes for high-risk systems and transparency, and how contracts along the AI value chain are handled.
What providers of general-purpose AI models must do under Chapter V: who counts as a provider, documentation, copyright and training-summary duties, the systemic-risk tier, the code of practice and its signatories, deadlines, and AI Office enforcement and fines.
How a high-risk AI system is shown to conform before it reaches the EU market: internal control versus notified body assessment, the route for Annex I products, certificates, reassessment, declaration, CE marking and registration, with the Commission's cost estimates.
How ISO/IEC 42001 AI management system certification relates to AI Act compliance: what the Act's presumption of conformity rests on, EN 18286:2026 as the harmonised quality management standard, the Commission JRC's view of the gaps, and where an AI management system still helps.
What AI Act compliance costs according to the Commission's own estimates: per-system compliance and conformity assessment costs from the 2021 impact assessment, human oversight costs for users, the savings claimed for the 2026 Digital Omnibus, SME relief and the fines that set the cost of non-compliance.
A step-by-step scope check for the AI Act: whether you use or supply an AI system or a general-purpose AI model, your role as provider, deployer, importer or distributor, reach to non-EU firms, the exclusions, and which risk tier and dates then apply.
How the AI Act reaches companies outside the EU: providers placing systems or models on the EU market, output used in the EU, authorised representatives for high-risk systems and general-purpose models, importer checks, and who enforces.
What Regulation (EU) 2026/1744 changed in the AI Act: the delayed high-risk dates and why, the ninth prohibition, the reworded AI literacy duty, simplified documentation and lower fines for small mid-caps, value chain cooperation, the AI Office's new powers and the Article 50 grace period.
How the 2024 Product Liability Directive adds civil liability on top of AI Act compliance: software and AI systems as products, the December 2026 dates, open-source carve-out, updates and continuous learning, presumptions of defect tied to safety and logging rules.