EU Code of Practice: No More Distinction Between AI-Generated and AI-Assisted Content

On March 5, 2026, the European Commission published the second draft of the Code of Practice for the marking and labeling of AI content. The most important change: The distinction between "fully AI-generated" and "AI-assisted" content has been completely removed.

This means: A photo retouched with AI and an image generated entirely by DALL-E now follow the same labeling rules.

What Has Changed?

The first draft (December 2025) proposed a two-tier taxonomy:

• "Fully AI-generated" – content created without human involvement
• "AI-assisted" – content with significant human contribution

The second draft has completely removed this distinction. There is now only one unified labeling regime for all content falling under Article 50 of the AI Act.

The 3 Major Changes at a Glance

1. One Label for Everything Whether your blog text was written by ChatGPT or merely edited with AI assistance – the same labeling obligation applies. This significantly simplifies compliance, as the difficult question "How much AI is in it?" is completely eliminated.

2. Open Standards Promoted The Code of Practice encourages the use of open standards for technical marking. This reduces costs and fosters interoperability between different tools and platforms.

3. Reduced Compliance Burden Consolidated measures, optional elements, and more flexibility to adapt solutions to each company's context.

The EU AI Icon Is Coming

The annex of the second draft includes for the first time illustrative examples of a European AI icon. A task force will develop the definitive symbol. The icon must be:

• Visible – from the first exposure to the content
• Consistently positioned – in a fixed, context-appropriate location
• Accessible – compliant with EU accessibility requirements
• Contain the abbreviation "AI" (or "KI" in German, "IA" in French)

What This Means for Website Operators

If you currently distinguish between different AI levels (e.g., "AI-generated" vs. "AI-assisted"), you should switch to a binary model:

• AI Content – any content where AI was involved
• No AI – purely human-created content

This simplification reduces compliance overhead and makes labeling clearer for users.

Key Timeline

| Date | Event | |------|-------| | March 5, 2026 | 2nd draft published | | March 30, 2026 | Feedback deadline passed | | June 2026 | Final version expected | | August 2, 2026 | Art. 50 transparency obligations legally binding |

How AIActify Helps

AIActify has already transitioned to the new system. Our snippet provides:

• Binary labeling – only "AI Content" or "No AI", compliant with the 2nd draft
• EU-compliant labels – visual AI badge with "AI"/"KI" identifier
• Schema.org metadata – machine-readable IPTC marking for crawlers
• Audit logs – immutable evidence for authorities

With a single script tag, make your website ready for the EU AI Act – in less than 2 minutes.

Conclusion

The 2nd draft of the Code of Practice sends a clear signal: simplicity over complexity. Dropping the AI/hybrid distinction is a pragmatic step that simplifies compliance. Companies should use the remaining three months until August 2026 to transition their systems.

Sources: EU Code of Practice – 2nd Draft, Regulation (EU) 2024/1689 Article 50