The EU Commission classifies ChatGPT as the first AI application under the strictest supervision category of the Digital Services Act, with Reddit and Roblox receiving the same status as particularly large platforms. The basis is user numbers of 159 million for ChatGPT and 57.2 and 48 million for Reddit and Roblox, respectively, in the EU. All three must submit risk assessments and audits within four months.
Commission bases classification on user numbers
The European Commission classifies ChatGPT as a “Very Large Online Search Engine” (VLOSE) because the chatbot conducts research on web queries and thus, according to the Commission, functionally corresponds to a search engine. This is the first such classification for an AI chatbot. Reddit and Roblox receive the status of “Very Large Online Platform” (VLOP). Both categories stem from the Digital Services Act, the European regulation for online services. A service is considered very large if it has 45 million average monthly users in the EU. OpenAI reported around 159 million monthly users for the ChatGPT search function for the period up to March 2026, Reddit 57.2 million, and Roblox nearly 48 million. The Irish media regulatory authority Coimisiún na Meán is responsible for overseeing ChatGPT and Reddit, while the Dutch competition authority ACM oversees Roblox. In total, 28 services now hold VLOP or VLOSE status, including Google Search and Microsoft Bing, which have held it longer as traditional search engines. ChatGPT has been growing rapidly in the EU for months, and since August the app has also shown advertising in its free tiers there – a sign of the reach that led to the classification.
Companies announce adjustments
OpenAI, Reddit, and Roblox have four months to demonstrate to the Commission sufficient precautions against risks to minors, elections, and public safety. This includes annual systemic risk assessments regarding illegal content, the protection of children and adolescents, the mental and physical health of users, fundamental rights, and electoral processes. The companies must also allow independent audits and share data with regulatory authorities and verified researchers so they can independently verify the risk reports. OpenAI states that the search function of ChatGPT already operates as a search service under the law, and that it is preparing for the additional obligations. Reddit says it has been preparing for the VLOP requirements for some time and is working with the authorities. Roblox emphasizes being the first gaming platform to reach this milestone and points to existing protective measures for underage users – a point that gains weight amid ongoing debates about youth protection features in chatbots. If a company fails to submit sufficient evidence by the deadline, the Commission can open a formal review procedure that may result in conditions or fines.
Fine framework based on previous DSA cases
The Commission can impose fines of up to six percent of global annual revenue for violations of the requirements. There is already a precedent: in May 2026, the authority fined Temu 200 million euros, and in July 2026 it fined AliExpress 550 million euros, in both cases for violations of DSA obligations for very large platforms. According to its own figures, the Commission has so far imposed around 870 million euros in DSA fines. The US government has repeatedly criticized earlier EU digital penalties against American technology companies as economic coercion and threatened countermeasures – a conflict that also resonates in the enforcement of the new requirements.
What matters now is whether the Commission actually backs up the deadline with its own reviews next year, or whether the companies’ self-disclosures go largely unchallenged. For ChatGPT, the classification is a test case: it shows that Brussels treats AI chatbots with web search the same way it treats traditional search engines under regulation – a pattern likely to catch up with other providers offering similar functions.
