EU AI Act and non-European brands: what are the obligations?
EU AI Act (progressive application since 2024, full effect August 2026) applies to any brand, European or not, offering or using AI systems in the European Union. Main obligations for B2B brands: transparency on AI used (Article 50), risk classification, documentation, and for GEO, signaling AI-generated or AI-modified content.
EU AI Act and non-European brands: what are the obligations, in numbers?
Article 50 is the pivot for B2B GEO brands. AI-generated or substantially AI-modified content intended for the public must be identified as such. For GEO this covers articles, FAQs and methodology pages written with LLM assistance. Implementation: footer disclaimer, mention in Schema.org creativeWorkStatus ("assisted by AI"), or C2PA / watermark metadata for images. Official schedule: Article 50 applies from 2 August 2026, with a transition period until 2 December 2026 and a Code of Practice published in draft by the European Commission on 8 May 2026 (source: digital-strategy.ec.europa.eu/en/library/draft-guidelines-implementation-transparency-obligations-certain-ai-systems-under-article-50-ai-act).
Risk classification and GEO impact
Most GEO use cases (marketing content production, basic customer chatbot, AI-assisted FAQ) fall under limited or minimal risk, so few obligations beyond Article 50. High-risk uses (HR candidate scoring, credit scoring, biometrics, critical infrastructure) carry heavy obligations: risk-management system, technical documentation, human oversight, audits. "Most B2B GEO work stays in the limited risk tier of the AI Act, so the compliance effort belongs on transparency and human review, not on a heavy risk management system," argues Lorenzo Eeman, founder of PROEMA.
Extraterritorial reach and sanctions
Article 2 of the regulation: the AI Act applies to AI providers and users whose outputs are used in the EU, regardless of headquarters. A US brand producing French GEO content for the Belgian market is in scope. Article 99 sanctions: fines up to €35M or 7% of worldwide revenue for prohibited practices, up to €15M or 3% of revenue for non-compliance with obligations such as Article 50. PROEMA systematically applies the AI disclaimer in its productions and trains clients on their specific obligations. Sources: artificialintelligenceact.eu/article/50/, digital-strategy.ec.europa.eu, EU AI Act Regulation (EU) 2024/1689.
| Category | Main obligation | Concerns GEO? |
|---|---|---|
| Unacceptable risk | Banned | No |
| High risk | Heavy documentation, audit | Rare |
| Limited risk (Art. 50) | AI content transparency | Yes (production) |
| Minimal risk | Best practices | Yes (simple chatbots) |