AI Labelling in TV Commercials: Advertisers Liable Starting August 2026
What Article 50 of the EU AI Act means for advertisers on EU broadcasters and why Switzerland is only seemingly off the hook
Article 50 of the EU AI Act stipulates that AI-generated or manipulated video content in TV commercials must be labelled. The advertiser alone is responsible for compliance and technical implementation on broadcasters such as RTL or ProSieben. Not the agency, not the broadcaster.
The rule is broader than many think. It applies not only to television, but also to social media.
Scope of Application
Mandatory from August 2026 for EU broadcasters such as ProSieben, RTL, M6, and TF1. Social media content is also covered.
Affected Content
Photos, videos, or audio that appear realistic and deceptively authentic in depicting people or places must be identifiable as AI-generated or manipulated.
Texts of Public Interest
Text-only social media posts also require labelling if they address topics of public interest and originate from AI without genuine human editorial review.
Exceptions
Purely artistic, satirical, creative, or fictional content may be exempt under certain circumstances, provided fraud or deception is excluded.
And Switzerland? These EU requirements do not apply to Swiss broadcasters. However, anyone running campaigns on EU broadcasters is still subject to these obligations. The advertiser is liable for quality and labelling, full stop.
My advice: do not rely on exceptions. Transparent labelling is also recommended in Switzerland. It costs little and avoids discussions if a commercial later crosses an EU border after all.
Conclusion: Adapt processes now, not when issues arise
The labelling obligation is here, and liability clearly rests with the advertiser. Anyone using AI in video production should firmly integrate labelling into the production workflow. It requires little effort. But it protects against a very real risk.