Transparency obligations (Article 50)
Chatbots and synthetic content.
Your mandated EU point of contact under Article 22 of the AI Act
Obligations for general-purpose AI models have applied since 2 August 2025. Providers established outside the EU must appoint an authorised representative in the Union under Article 54 - a separate mandate from Article 22 for high-risk systems.
Providers of models released under a free and open-source licence are exempt from the first two and from Article 54, unless the model presents systemic risk.
A model is presumed to have high-impact capabilities when the cumulative compute used for training exceeds 10^25 floating-point operations. Providers of such models must also perform model evaluations including adversarial testing, assess and mitigate systemic risks, report serious incidents to the AI Office, and ensure adequate cybersecurity.
| Model | Obligations apply |
|---|---|
| Placed on the market from 2 August 2025 | Immediately |
| Placed on the market before 2 August 2025 | By 2 August 2027 |
The General-Purpose AI Code of Practice can be used to demonstrate compliance. Under Regulation (EU) 2026/1744, the AI Office also has exclusive competence over AI systems built on a GPAI model by the same provider or group, with exceptions such as AI in Annex I products (Article 75(1)).
Yes. Before placing a general-purpose AI model on the EU market, a provider established in a third country must appoint an authorised representative in the Union by written mandate (Article 54), unless an open-source exemption applies.
It is presumed when training compute exceeds 10^25 floating-point operations, or the Commission designates it based on the criteria in Annex XIII.
No. Article 54 covers general-purpose AI models and cooperation with the AI Office; Article 22 covers high-risk AI systems. One company can hold both mandates, but they are separate.
Chatbots and synthetic content.
Fines for GPAI providers.
The high-risk AR role.
Note: This guide provides general information about Regulation (EU) 2024/1689 (AI Act) as amended by Regulation (EU) 2026/1744 (Digital Omnibus on AI), not legal advice. Last reviewed 5 October 2026.