EU labelling rules for AI-generated sports content: what Norwegian sportstech needs to know
From 2 December, providers of older generative AI systems must also meet EU requirements for machine-readable marking of synthetic content. The rules affect a sports market using AI for match reports, video production, commentator voices and digital media experiences. For Norwegian technology suppliers operating in the EU market, the question is which obligations apply to their particular products and services.
<p><strong>From 2 December, providers of older generative AI systems must also meet EU requirements for machine-readable marking of synthetic content. The rules affect a sports market using AI for match reports, video production, commentator voices and digital media experiences. For Norwegian technology suppliers operating in the EU market, the question is which obligations apply to their particular products and services.</strong></p> <p>An automated highlights package, a synthetic commentator and an AI-written match report can form part of the same production. Yet they are not necessarily subject to the same labelling obligations. Companies must distinguish technical marking by the system from information audiences must receive.</p> <h2>The December deadline is not the start of the rules</h2> <p>The <a href="https://ai-act-service-desk.ec.europa.eu/en/ai-act/eu-ai-act-implementation-timeline">European Commission’s timeline</a> confirms that Article 50 applies from 2 August 2026. The transparency rules are therefore already applicable in the EU.</p> <p>However, <a href="https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32026R1744">Regulation (EU) 2026/1744</a> introduced a limited transition. Under <a href="https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-111">Article 111(4)</a>, providers of systems generating synthetic audio, images, video or text that were placed on the market before 2 August must comply with Article 50(2) by 2 December 2026.</p> <p>The transition covers this technical marking and detection obligation, not all transparency rules. Systems placed on the market on or after 2 August do not receive this extension. December is therefore a deadline for certain existing systems, not a general starting date for AI-content labelling.</p> <h2>Two obligations that must not be confused</h2> <p>A provider develops or has an AI system developed and places it on the market or puts it into service under its own name. A professional user, called a “deployer”, uses the system under its authority. One business can hold both roles.</p> <p><a href="https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50">Article 50(2)</a> requires providers of relevant generative systems to ensure machine-readable marking and detectability of synthetic content. Solutions must be effective, interoperable, robust and reliable as far as technically feasible, taking account of factors including content type, costs and the state of the art.</p> <p>This differs from a visible label on all AI-assisted content. The provision includes exceptions for standard editing assistance and where a system does not substantially alter the user’s input data or its meaning.</p> <p>Article 50(4) covers, among other things, professional users’ disclosure of deepfakes: AI-generated or manipulated audio, images or video resembling existing people, places or events and falsely appearing authentic. Audiences must be able to perceive the disclosure; hidden technical marking is insufficient. Evidently artistic or fictional works have a tailored disclosure obligation, not a blanket exemption.</p> <p>AI-generated text published to inform the public about matters of public interest has a separate exception: it must have undergone human review or editorial control, and a person or organisation must hold editorial responsibility. The <a href="https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act">Commission’s explanation</a> makes clear that mere spell-checking is insufficient.</p> <h2>Four sports products, four different assessments</h2> <p><strong>Automated match reports:</strong> The provider must examine whether the text system falls within technical marking requirements. The publisher must separately assess public interest and disclosure. Who checks the facts and substance, and who holds editorial responsibility? This determines whether the text exception may be relevant.</p> <p><strong>Automated highlights:</strong> Finding and cutting authentic match footage does not automatically create a deepfake. The question is whether the tool merely edits, or creates or substantially manipulates content. The technology and output need assessment, not just the “AI” marketing label.</p> <p><strong>Synthetic commentary:</strong> Generated speech may trigger the provider’s technical marking obligation. A separate disclosure duty depends, among other things, on whether the audio qualifies as a deepfake. Does the voice imitate a real commentator and falsely appear authentic, or is it clearly artificial?</p> <p><strong>Sponsor content:</strong> Realistic synthetic video of an athlete or sporting event may require deepfake disclosure. Companies should examine the impression of authenticity and <a href="https://digital-strategy.ec.europa.eu/en/policies/guidelines-ai-transparency-obligations">relevant exceptions and adaptations</a>, alongside consent, privacy, rights and misleading advertising. An AI label does not replace these checks.</p> <h2>Norwegian legislation and European market access</h2> <p>The <a href="https://www.regjeringen.no/no/aktuelt/tung-vil-sende-ki-loven-med-endringer-pa-horing/id3169693/">Norwegian government announced on 4 August</a> a new consultation on its AI bill in autumn 2026, with the ambition of presenting legislation to parliament in spring 2027. The <a href="https://www.regjeringen.no/no/dokumenter/3112327/id3112327">first consultation, from 2025</a>, is still listed as under consideration.</p> <p>In source checks on 22 October, NST has not found a published new consultation on regjeringen.no. The August announcement documents the plan, not that the new consultation has actually opened. This remains an unresolved status point, not confirmation that the consultation has been cancelled.</p> <p>The AI Act does not automatically become Norwegian domestic law before the relevant EEA incorporation and implementation. Norwegian companies may nevertheless fall within the <a href="https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-2">EU regulation’s territorial scope</a>, for example when placing systems on the EU market or where their outputs are used in the EU. Having an EU customer does not automatically make every provision applicable: role, product and use require individual assessment.</p> <p>Nor is Norway a legal vacuum. Privacy, copyright and marketing rules already apply.</p> <p>For companies selling to clubs, leagues, broadcasters and rights holders, this affects product development, technical architecture, customer responsibilities and contracts. Documentation and operation may bring costs that need to be included in commercialisation. NST’s coverage of <a href="/news/forzasys-ai-orchestrator-sportsinnhold-simula">AI-enabled sports production</a> and the <a href="/news/ioc-ki-sjekkliste-norsk-idrett-2026">IOC’s AI checklist</a> provides sector context, not documented violations or assessments of named companies’ compliance.</p> <h2>Four questions before the next product delivery</h2> <ol> <li>Are we the AI-system provider, the professional user, or both?</li> <li>Does our product generate synthetic content covered by Article 50(2)?</li> <li>Does our use create a separate disclosure obligation under Article 50(4)?</li> <li>Which technical and contractual measures must document compliance when we operate in the EU market?</li> </ol> <p>These are questions for a specific assessment, not a single legal answer for the entire industry. Transparency is becoming part of product development: understanding the difference between technical marking and audience-facing disclosure is also part of building sports media services that customers can trust and buy across markets.</p> <p><em>Illustration: A person using a laptop at a football stadium at sunset. The image does not document AI-generated content or a regulatory breach.</em></p>