In brief: Since August 2, 2026, companies are only required to label deepfakes, direct AI customer communication, and unedited AI-generated texts of public interest – most marketing and business applications remain unaffected.
Since August 2, 2026, the EU AI Act has established binding transparency obligations for the first time for certain AI-generated content. Those affected are primarily deepfakes and direct communication with AI systems – but not every use of AI within a company.
The EU AI Act provides for binding transparency obligations for AI-generated content starting August 2, 2026. The aim is to protect consumers and business partners from misleading representations and to make the use of artificial intelligence more traceable. Contrary to what is often assumed, however, this is not a blanket labeling requirement for all AI-generated content, but rather requirements tied to clearly defined use cases.
Accordingly, labeling is required for AI-generated or manipulated images, videos, and audio recordings that deceptively depict real people, objects, places, or events as authentic – so-called deepfakes. Companies must likewise disclose when customers are communicating directly with an artificial intelligence, for example via chatbots or virtual assistants on websites and in customer portals. A stricter condition applies to AI-generated texts: labeling is generally only required if the text is published without human review and serves to inform the public on matters of public interest.
For compliance practice, this means that many typical business applications remain unaffected by the labeling obligation. Those who use AI to create product descriptions, marketing texts, translations, or internal documents, or to analyze business data, generally do not need to label such content. The same applies to mere linguistic polishing or automated spelling correction of existing texts. What matters is always whether AI-generated content is editorially reviewed by a human before publication, or whether a person assumes editorial responsibility and liability for the published text – in this case, the specific transparency obligations do not apply.
Even for audiovisual media, not every AI-assisted edit automatically triggers a labeling obligation. Those who merely extend the image background, optimize colors, improve image quality, or remove disruptive elements generally fall outside the scope of the regulations. Nevertheless, companies should review their processes and internal responsibilities to determine which content falls within scope, who bears editorial responsibility, and how any required labeling is to be implemented technically and organizationally.
Source: www.it-daily.net · Published August 14, 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.8.3.