Do I Have to Label AI-Written Product Descriptions?

Short answer: usually not, but it depends on what the text does. The EU AI Act, Regulation (EU) 2024/1689, includes an Article 50 duty to disclose AI-generated or manipulated text, but that duty is framed around text that informs the public on matters of public interest. An ordinary product description, telling a shopper about a jumper’s fabric and fit, is not normally that kind of public-interest text, so the strict marking obligation often will not apply.

When does labelling matter more?

The picture changes as content moves away from plain catalogue copy. As a deployer of third-party AI, you carry Article 50 duties, and some shop content clearly falls within them:

• AI-generated product images should be marked as artificially generated.
• Chatbots must tell customers they are interacting with AI, not a person.
• Deep fakes and manipulated media must be disclosed.

So while the AI-written words in a description often escape the strict text-marking rule, the AI-made photo next to them and the AI chatbot answering questions about them frequently do not.

What is the sensible approach?

Even where labelling is not strictly required, keep AI-drafted descriptions accurate and review them before publishing. Misleading claims can trigger consumer-protection rules regardless of who or what wrote them. If a description drifts into information the public relies on, or you are simply unsure, a brief disclosure is a low-cost way to stay on the safe side.

For more on this, see our guide to the complete AI Act guide and our FAQ on AI transparency for online stores.

This answer is educational and does not constitute legal advice.