„From August 2, all AI content and chatbots must be labeled; companies are unprepared, and incorrect labeling can result in fines in the billions.”
Counter-narrative
Not all AI-generated content needs to be labeled automatically, as rules vary by content type and labeling alone does not solve everything. This is represented by a Hungarian media lawyer and materials analyzing transparency rules.
Carriers
Hungarian professional and business media (Economx, Blikk series), consultants and SME checklists, Artificial Intelligence Office information page., AI providers' watermarking announcements, Dutch survey on chatbot labeling
Analysis
The narrative is alarm tied to the moment the rules enter into force, on which compliance guides and advisory materials are built. Three things feed it: the August deadline, the fine of up to €15 million (described in the Hungarian press as running into the billions), and providers' watermarking announcements. The support is weak. The procedural rules in force from August cover the evaluation of general-purpose models and fines on them. OpenAI itself says text watermarking and detection are still technologically immature, so it is introducing them in phases.
Evidence: weak — The fine threat stems from the statutory ceiling (up to €15 million) and from press interpretation. The stories report no fine imposed under Article 50. Unpreparedness is measured by only one Dutch survey; the claim about Hungarian companies (Economx) is not based on measurement.
What's at stake: Whether compliance resources are directed at real risks or toward conspicuous labeling.
Decisions on what to label must be made by content type: realistic images, audio, and video, public interest text, and chatbot interactions. Labeling everything is not the goal. For chatbots, the AI nature must be communicated at the point of interaction in the user's language.
Under the AI Act, users must be informed when interacting with a chatbot or encountering deepfakes.
Hungary
In Hungary, labeling was a frequently discussed topic during the period under review. The Artificial Intelligence Office launched an information page, and according to Economx, most Hungarian companies still do not know what to prepare for (2026-08-03-08). Trade media debated readiness and what actually needs to be labeled (2026-08-10-05). KPMG materials, an SME checklist and a Blikk series appeared (2026-08-24-16), along with an SME guide (2026-10-05-60). A media lawyer said that not everything needs to be labeled (2026-08-31-28). Mandatory labeling of synthetic content was also a topic at the hearings of candidates for president of the National Media and Infocommunications Authority (NMHH) (2026-09-14-37). The stories report no fines in Hungary so far.