Three parts of the EU AI Act every marketer should know
The Act does not regulate advertising directly, but it reaches into your workflow.
Background
The EU AI Act does not regulate advertising as a category, but several of its provisions reach directly into advertising and media buying workflows. From 2 August 2026, transparency rules are in force and enforcement has begun.
Three key points
1. Label synthetic content. If you use generative AI to create or alter image, audio or video content in a way that could pass for real, you must visibly disclose it. Generative AI providers must embed machine-readable marks in outputs. The Commission has published guidelines and created icons to support labelling. A voluntary Code of Practice, with around 234 signatories by late September 2026, offers a recognised route to demonstrating compliance.
2. Do not manipulate. The Act prohibits AI that manipulates people, uses subliminal techniques or exploits emotional or psychological vulnerabilities. The Commission's guidelines on prohibited practices include worked examples. Personalisation and targeting tools that cross this line are at risk regardless of how they are packaged.
3. Watch recruitment advertising. The Commission classifies AI used to place targeted job advertisements as a high-risk employment use. Full compliance obligations for these systems apply from December 2027, but the classification affects how such tools should be documented and monitored now.
Why this matters
The duty to label is not sector-specific: it applies to any deployer producing AI-generated content that could mislead. Marketing teams that have not yet agreed an internal labelling policy, or checked whether their agency and production partners are compliant, need to move quickly. Enforcement is active.
Sources: EU AI Act, EUR-Lex; Commission Article 50 guidelines; Code of Practice on AI-generated content.
