Artificial Intelligence Act (AI Act)

Dates

• 2024-08-01: Entry into force (OJ 12-7-2024) Official source
• 2025-02-02: Chapters I and II apply: AI literacy (Art. 4) and prohibited practices (Art. 5) Official source
• 2025-08-02: General-purpose AI model obligations apply (Chapter V) Official source
• 2026-08-02: General date of application, including the Art. 50 transparency obligations Official source
• 2026-12-02: The new prohibitions of Art. 5(1)(ba) and (bb) and Art. 5(1a) and (1b) apply (Art. 113, third paragraph, point (a)); end of the Art. 50(2) transition for generative systems placed on the market before 2-8-2026 (Art. 111(4)) Official source
• 2027-12-02: Annex III high-risk requirements apply Official source
• 2028-08-02: Annex I (product) high-risk requirements apply Official source

What it means for your store

• Obligation

  If you use AI systems in your business (for example, a support assistant or a tool that handles agent orders), you must take measures to support the AI literacy of your staff and of those operating them on your behalf. No specific level must be guaranteed.

  Art. 4(1) (redacción del Reglamento (UE) 2026/1744) · Official source

• Obligation

  If you put into service under your own name an AI system that converses with buyers (your own shopping assistant or agent, developed by you or for you), you are a provider and must inform people they are interacting with AI, unless obvious, at the latest at the first interaction. Applicable from 2-8-2026.

  Arts. 3(3), 50(1) y 50(5) · Official source

• Obligation

  If you yourself use an AI system to generate or manipulate images, audio or video resembling real persons, objects, places or events that could pass as authentic (deep fakes), you must disclose they are artificial. An AI-generated product photo that could pass as real probably qualifies. Exceptions: use authorised by law for criminal-justice purposes; for evidently artistic, creative, satirical or fictional works, the obligation is limited to disclosing the existence of such content in an appropriate manner that does not hamper the display or enjoyment of the work. Applicable from 2-8-2026.

  Arts. 3(60) y 50(4) · Official source

• Obligation

  If you provide a system generating synthetic text, image, audio or video, its outputs must be marked in a machine-readable format. For systems placed on the market before 2-8-2026, the deadline is 2-12-2026.

  Arts. 50(2) y 111(4) · Official source

• Obligation

  If you use AI to evaluate the creditworthiness or credit score of natural persons (for example, to grant deferred payment), from 2-12-2027 you will have the obligations of a deployer of a high-risk system, including the fundamental rights impact assessment. AI used to detect financial fraud is excluded from this category.

  Anexo III, punto 5(b); arts. 26 y 27(1); art. 113, párrafo tercero, letra c) · Official source

Where to start

• Inventory the AI systems you use or put into service under your name and record your role for each (provider or deployer): your obligations depend on that, not on agents buying from you.
• Level 3: if you expose your own assistant or agent to buyers, show the AI notice at the first interaction.
• Level 2: if you use AI-generated product images that could pass as real photos, label them as AI-generated.
• Level 5: an anti-fraud system screening agent orders is not high-risk by itself; if it also decides on credit or deferred payment for natural persons, prepare the impact assessment before 2-12-2027.

Related levels

Level 2: Understandable; Level 3: Operable; Level 5: Trustworthy