News
EU to Require Labeling of AI-Generated Content Starting August 2
The European Commission has published guidelines for the code of practice on labeling AI-generated content, which takes effect August 2, 2026 under Article 50 of the AI Act. Contrary to earlier fears, the requirement won't cover every AI-assisted SEO text or product description.
Contents
Starting August 2, 2026, Article 50 of the AI Act will take effect across the European Union, requiring labels on content generated by artificial intelligence. The European Commission has published guidelines for the accompanying code of practice, spelling out exactly what needs to be labeled and what falls outside the requirement.
What must be labeled
The regulation primarily targets deepfakes, meaning images, videos and audio recordings generated or substantially altered by AI in ways that resemble real people, places, products or events. In practice this covers things like virtual models showcasing clothing in ads, synthetic photo shoots, AI-generated voices impersonating real people, and fake customer testimonials. Labeling is also required for AI-generated or manipulated text published on matters of public interest, unless it has undergone editorial review by a human who takes responsibility for it.
What falls outside the requirement
Contrary to earlier fears from the e-commerce industry, standard product descriptions, category content and SEO articles created with the help of generative AI will not automatically require labeling just because AI tools were used. The exemption applies as long as the content goes through editorial review by a human who formally takes responsibility for its publication. The distinction is meant to prevent a situation where every online store would have to label thousands of product descriptions.
Technical requirements for providers
Providers of generative systems are required to implement machine-readable marking based on metadata, digital signatures and Content Credentials-type solutions that allow the origin of a piece of content and its modification history to be traced. The markers are meant to be resistant to removal and tailored to the type of content, whether image, audio, video or text. The code of practice was developed with input from more than 180 stakeholders, including AI providers, small and medium-sized businesses, researchers and civil society organizations, coordinated by the EU's AI Office.
Signing the code of practice itself is voluntary, but the obligations under Article 50 of the AI Act are legally binding regardless of whether a company signs the document. Signatories only gain administrative convenience, since the Commission considers their implementation approach compliant with the law without additional checks. The AI Act's general penalty range reaches up to 35 million euros or 7 percent of a company's global turnover for the most serious violations, though the specific practice of enforcing content-labeling rules has yet to take shape.
What it means for Poland
The new EU requirements coincide with work in the Sejm (the lower house of Poland's parliament) on a national law on artificial intelligence systems, which will establish a Polish supervisory authority responsible, among other things, for enforcing the AI Act within the country. Companies operating in Poland, especially marketing agencies, online stores using AI-generated product images, and publishers running AI-assisted journalistic content, should already be checking their processes against the August deadline.
In practice, this means reviewing internal content-approval procedures to clearly document which materials undergo human editorial review and which are published without it. Lacking such documentation could make it harder to demonstrate compliance during an inspection, even if the content itself raises no concerns about accuracy.
Enforcement will also run into practical difficulties, since there is no widely accepted technical test yet that can definitively determine whether a given image or recording was generated by AI. The Commission is counting on the shared labeling standards developed under the code of practice to gradually make it easier to automatically detect such content across the EU.
Sources: AI Act od sierpnia 2026. Które treści AI trzeba będzie oznaczać (ewp.pl), Unijny kodeks oznaczania treści AI. Od sierpnia nowe obowiązki (cyberdefence24.pl), Od sierpnia nowe obowiązki dla treści AI (ecommercenews.pl), Code of Practice on marking and labelling of AI-generated content (digital-strategy.ec.europa.eu).
