EU AI Act requires labels for realistic AI ads

From August 2, 2026, brands and advertisers whose campaigns reach people in the European Union must disclose AI-generated or AI-altered images, video, or audio when the material could pass for real. The obligation applies based on where the ad is seen, not where the company is based, so US brands running AI-built creative in Germany, France, or Spain fall inside the rule.

The EU AI Act assigns one task to software providers and another to publishers. AI tools are expected to add hidden markers to generated or altered files, while advertisers must provide a clear, visible notice at the moment an audience first encounters realistic AI content. Ordinary retouching, dust removal, color adjustments, cropping, and minor cleanup of real photos are treated differently from fabricated people, invented scenes, or realistic location swaps.

The labeling provision starts on schedule even though other high-risk AI rules were delayed. Labeling breaches can trigger penalties of up to 15 million euros or 3% of total worldwide turnover, whichever is higher, with smaller companies capped at the lower of the two figures. The practical compliance steps are to audit EU-facing creative, keep tool-generated markers, standardize labels, add final-review checks, and label gray-area work.