ChatGPT Content on Your Website: Do You Need to Label It?
More and more companies use ChatGPT for blog posts, product descriptions, and FAQs. But from 2026: those who don't label risk heavy fines.
Christian Lechner
ChatGPT has revolutionized the way companies create content. Product descriptions, blog posts, FAQ pages, social media posts – AI is used everywhere. But what many don't know: From August 2026, this content must be labeled.
What Does the EU AI Act Say About AI Text?
Article 50(2) of the AI Regulation is clear: Providers of AI systems that generate synthetic text must ensure that the output is marked as machine-generated. This directly affects anyone using ChatGPT, Claude, Gemini, or similar tools for their website.
The labeling must be machine-readable – a simple note saying "written with ChatGPT" is not sufficient.
Which Texts Must Be Labeled?
- Fully AI-generated texts: Blog posts copied 1:1 from ChatGPT
- AI-assisted texts: Content where AI contributed significant parts
- Product descriptions: When created with AI tools
- FAQ answers: Automatically generated help texts
Exception: Texts that were only edited with AI spell-checkers or grammar tools do not fall under the labeling requirement.
How to Label Correctly?
The EU requires two levels of labeling:
1. Visible Labeling A visual label like "AI-generated" or "AI-assisted" must be visible to the user. AIActify offers configurable badges that automatically appear on marked content.
2. Machine-Readable Metadata
Schema.org JSON-LD with the isBasedOn attribute or IPTC digitalSourceType: trainedAlgorithmicMedia must be embedded in the HTML code. AIActify injects this metadata automatically.
The Easiest Way: AIActify Visual Editor
With the On-Site Visual Editor from AIActify, you can mark which texts are AI-generated directly on your website with a simple click. No code, no effort – and still 100% legally compliant.
Ready for the EU AI Act?
Set up in 2 minutes. Free scanner & 15-day trial.