EU Designates ChatGPT a Very Large Online Search Engine

The European Commission has designated ChatGPT a Very Large Online Search Engine under the Digital Services Act, and OpenAI has until the end of December 2026 to comply.

The designation, reported by The Verge, puts ChatGPT in the same regulatory tier as the largest platforms operating in Europe. Reddit and Roblox were named in the same announcement, classified as Very Large Online Platforms and subject to the same rules.

What the designation actually obliges OpenAI to do

Per The Verge's report, OpenAI "will soon be held accountable for mitigating risks related to ChatGPT's impact on minors, user mental health, and the spread of illegal content in the European Union." Those three risk categories are the substance of the change — everything else follows from them.

The DSA's concrete restrictions are narrower than the risk language suggests. The report lists two: platforms may not target ads to minors, and may not target using a person's sexual orientation, religion, ethnicity, or political beliefs. Separately, they must provide more transparency about how their recommendation algorithms work.

The threshold for the designation is a user count, not a capability assessment. Under the DSA, online platforms and search engines are considered "Very Large" once they reach at least 45 million average monthly users in the EU. That is the whole test. A chatbot crosses it the same way a social network does.

Henna Virkkunen, EU executive vice-president for tech sovereignty, security and democracy, framed it in the announcement: "These new designations mean that ChatGPT, Reddit and Roblox will now be held to a higher standard of scrutiny and accountability in the European Union, in line with their large impact on our citizens and society."

Search engine, not platform — and why that distinction matters

ChatGPT was classified as a Very Large Online Search Engine. Reddit and Roblox were classified as Very Large Online Platforms. The Commission drew a line, and it put a chat interface on the search side of it.

A developer writing on Dev.to read the same signal: the classification "signals that AI products that retrieve, rank, summarize, and surface information from the web are increasingly being treated as part of the search ecosystem." That is a reasonable inference from the category the Commission chose, though the Commission's own reasoning for the split was not detailed in the reporting.

The same post argues the boundary itself is dissolving: "Search is no longer limited to a traditional search box. AI systems can interpret a question, retrieve information, rank sources, synthesize answers, and present a response without requiring users to browse dozens of pages." Regulators now have to fit that behaviour into categories written before it existed, and the VLOSE label is the first attempt.

What this means if you ship an AI product in Europe

Nothing in the designation applies to you directly unless you clear 45 million monthly EU users. The second-order effects are the ones worth planning around.

Transparency work moves upstream. If a designated service must explain how it ranks and recommends, the components it depends on — retrieval layers, reranking models, citation logic — need to produce explanations too. Teams building on the OpenAI API should expect more questions about provenance, not fewer.

The "AI assistant" label stops being a shelter. The precedent here is that regulators will classify by what a product does, not what it calls itself. A retrieval-augmented assistant that ranks web sources and synthesizes an answer is doing search, and one regulator has now said so in a formal designation.

Content and minors policy becomes a product requirement, not a policy page. Two of the three named risk areas — impact on minors and user mental health — are behavioural, not technical. They get addressed in interaction design, refusal behaviour, and age handling, which are product decisions made by engineers.

The Dev.to post lists the open questions the developer community now inherits, and they are the right ones: how AI search systems should handle illegal or harmful content, how transparent AI-generated results should be, what responsibility developers carry when a system decides which information users see, and how the regulation will reach APIs, agents, browsers, and developer tools.

What to watch between now and December

Three things will tell you how much this designation actually changes.

First, whether OpenAI ships EU-specific behaviour in ChatGPT before the deadline, or argues its existing safeguards already satisfy the obligations. The compliance date is the end of December 2026, which leaves four months.

Second, whether other assistants get designated. The 45-million threshold is objective and public, so the question is which services cross it and when — and whether the Commission classifies them as search engines too, or splits the category.

Third, whether other jurisdictions copy the classification. The Dev.to post raises this question directly, and it is unresolved. Nothing in the reporting indicates another regulator has adopted the classification.

For now, the practical takeaway is narrow and firm: an AI assistant with enough European users is a search engine in the eyes of EU law, and the compliance clock started this week.