New European Union regulations require companies to label chatbots, deepfake content, and AI-generated marketing material, bringing a mandatory transparency obligation to the consumer-facing output layer. The framework draws an explicit comparison to cookie consent notices: a disclosure mechanism users encounter before or alongside AI-generated content, regardless of what generates it underneath.
Where the compliance obligation sits in the stack
The cookie banner parallel is more than rhetorical. Cookie consent requirements under GDPR attached to the data-collection layer of web products, forcing a disclosure at the moment of user interaction rather than at the infrastructure level. EU AI labeling rules operate on the same logic: the obligation attaches to output, whether that output is a chatbot response, a synthesized video, or generated copy in a marketing campaign.
That distinction shapes how companies have to think about compliance. A single underlying model can produce labeled chatbot interactions in one consumer-facing product and unlabeled internal summaries in another. The scope of the obligation turns on what reaches a consumer, not on which systems a company operates. This is a content-classification problem as much as a technical one.
Three content types, one shared risk
The regulation covers chatbots, deepfakes, and AI-generated marketing material. The common thread across all three is the potential for a consumer to mistake AI-generated content for human-produced content. Chatbots must identify themselves as automated systems. Deepfakes, synthetic media that replaces or fabricates human likenesses, require disclosure of their artificial origin. Generated marketing material carries the same requirement.
The marketing category is the broadest and least clearly bounded. Campaigns increasingly blend generated text with human editing, and the regulation as described does not draw a line on where the threshold for disclosure falls in a hybrid workflow.
The cookie banner precedent
GDPR cookie consent notices became a near-universal feature of the consumer web after enforcement began. They added a disclosure layer on top of behavioral advertising without dismantling it. The EU's AI labeling framework appears to follow the same architecture: compliance-by-disclosure rather than compliance-by-restriction. Whether companies treat the requirement as a meaningful label or an easily dismissed banner is a product design question the regulation, as summarized, leaves unanswered. The marketing material category, as the broadest of the three, is where that question will likely surface first.