For product content teams, the EU AI Act (Regulation (EU) 2024/1689) mainly means Article 50 transparency duties, applicable since 2 August 2026, and an obligation to support AI literacy. Human-reviewed product descriptions generally need no AI label, but shopping chatbots must disclose that they are AI and realistic synthetic images may count as deepfakes.
For product content teams, the EU AI Act (Regulation (EU) 2024/1689) mainly brings two obligations: the Article 50 transparency duties, which have applied since 2 August 2026, and the Article 4 duty to support AI literacy among staff. Human-reviewed product descriptions generally need no AI label. Customer-facing shopping assistants must disclose that they are AI, and realistic AI-generated product imagery needs assessment under the deepfake rules.
How the AI Act applies to product content teams
The AI Act assigns duties by role. Most brands, manufacturers and retailers using AI for product content are deployers: they use AI systems built by others, such as AI features in a PIM, a copywriting tool or a general-purpose model accessed through an API. Software vendors that place AI systems on the market are providers and carry most technical obligations, including machine-readable marking of synthetic output.
An organisation can become a provider when it develops its own AI system or places one on the market under its own name. Companies building in-house generation pipelines should assess that question with legal counsel.
Provider duties still matter to deployers during software selection. Buyers can ask PIM, DAM and syndication vendors how their AI features meet Article 50(2) marking, which models they use and whether customer data is used for training, as part of wider PIM selection criteria.
What Article 50 requires and when
| Paragraph | Who | Obligation | Product content relevance |
|---|---|---|---|
| 50(1) | Providers | Systems interacting with people must inform them they are dealing with AI, unless obvious | Shopping chatbots and product advisers |
| 50(2) | Providers | Mark synthetic audio, image, video and text in a machine-readable, detectable format | Built into generation tools by vendors |
| 50(3) | Deployers | Inform people exposed to emotion recognition or biometric categorisation | Rare in product content |
| 50(4) | Deployers | Disclose deepfakes, and AI text published to inform the public on matters of public interest | Synthetic imagery, editorial content |
The Digital Omnibus on AI, published in the Official Journal on 24 July 2026 and in force since 27 July 2026, kept 2 August 2026 as the Article 50 start date. It added a grace period to 2 December 2026 for machine-readable marking by generative systems placed on the market before 2 August 2026.
The Commission published a voluntary Code of Practice on marking and labelling AI-generated content on 10 June 2026. It sets out a multi-layered marking approach for providers, such as metadata combined with watermarks, and label placement guidance for deployers, including an EU icon.
Do AI product descriptions and images need a label
Text: product descriptions are commercial information, and the Article 50(4) text duty targets text published to inform the public on matters of public interest. The duty also does not apply where text has undergone human review or editorial control and a person holds editorial responsibility. A documented review workflow therefore matters, both for accuracy and as evidence of editorial control.
Images and video: the AI Act defines a deepfake as AI-generated or manipulated content that resembles existing persons, objects, places or events and would falsely appear to a person to be genuine or truthful. A photorealistic lifestyle scene showing a real product in a setting that never existed may meet that definition. Background removal or colour correction is less likely to. Content that is evidently creative or fictional benefits from lighter disclosure requirements. Teams should classify their image use cases and decide labelling rules per case.
Assistants: a product adviser or chatbot on a shop must make clear that the customer is interacting with AI, unless this is obvious from the context.
Consumer law applies alongside the AI Act. Misleading AI-generated claims remain prohibited under the Unfair Commercial Practices Directive.
What else changed with the Digital Omnibus on AI
- High-risk systems: Annex III obligations moved from 2 August 2026 to 2 December 2027, and Annex I (AI in regulated products) to 2 August 2028.
- AI literacy: Article 4 now requires providers and deployers to take measures supporting AI literacy, without guaranteeing a specific level per person. Documented training is the expected approach.
- Other dates: general-purpose AI model obligations have applied since 2 August 2025, and prohibitions since 2 February 2025.
Fines for breaching Article 50 reach up to €15 million or 3% of worldwide annual turnover, whichever is higher.
Practical next steps
- Inventory every AI use in product content: text, translation, classification, images, video and assistants.
- Determine the organisation's role (deployer or provider) for each system.
- Document human review and editorial responsibility for AI-assisted product text.
- Classify synthetic imagery and set disclosure rules for realistic scenes.
- Confirm that customer-facing assistants disclose their AI nature.
- Record AI literacy training and add AI provenance fields under product data governance.
Frequently asked questions
Do AI-generated product descriptions need to be labelled under the EU AI Act?
Generally not. The Article 50(4) text duty covers AI-generated text published to inform the public on matters of public interest, and it does not apply where text has undergone human review or editorial control with a person taking editorial responsibility.
Are e-commerce AI systems high-risk under the AI Act?
Typical product content uses, such as copywriting, translation, classification and image editing, are not listed as high-risk in Annex III. High-risk rules for Annex III systems now apply from 2 December 2027 following the Digital Omnibus on AI.
What are the penalties for breaching Article 50?
Non-compliance with Article 50 transparency obligations can lead to fines of up to EUR 15 million or 3% of worldwide annual turnover, whichever is higher, with lower caps for SMEs.