The Search Engine That Answers

The question of what to call a chatbot just got answered by the only institution that gets to answer such things with binding force. On August 31, the European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act — the VLOSE tier, reserved for services exceeding 45 million average monthly EU users, carrying the regime’s heaviest obligations. Reddit and Roblox were designated the same day. OpenAI’s assistant is now in the same regulatory category as Google Search, and it has four months to comply.

The reasoning, per the Commission: ChatGPT can search the web and synthesize what it finds, which makes it a hybrid service, and the hybrid got sorted into the search bucket. Whether that is a lawyerly stretch or an act of precision depends on your view of what a search engine is. The Commission’s view is the one that costs money.

Why the Category Matters

The DSA does not treat platforms and search engines identically, and the VLOSE tier is not the VLOP tier. Search engines face their own stack of obligations — distinct risk profiles, distinct mitigation duties. Dropping a generative assistant into the search category does two things at once. First, it subjects ChatGPT to the full systemic-risk apparatus: assessments and mitigation of illegal content, minors’ safety, mental health, fundamental rights, electoral integrity, public security. Second, it sets the taxonomy for every generative service that follows. If ChatGPT is a search engine, what is Gemini? What is the next assistant with browsing? Brussels has just written the default answer, and defaults in regulation are forever.

The four-month clock is the practical headline. By late December, OpenAI must have completed and published risk assessments, stood up mitigation machinery, and opened the transparency plumbing the DSA demands of the largest players. Users in the EU will see the difference first in disclosure: ads labeling where applicable, recommender transparency, researcher data access. Behind the scenes, it is a compliance organization being built at startup speed.

The Enforcement Precedent

The Commission has shown it will use the DSA’s teeth — X has already felt formal proceedings, and the VLOSE tier carries penalties up to six percent of global turnover. For OpenAI, a company that has spent two years positioning itself as the responsible incumbent, the designation is an odd kind of recognition: too big to be a startup, too consequential to self-certify. For the EU, it closes an obvious gap before the next election cycle rather than after — generative assistants are already routing citizens to information, and the DSA’s authors were never going to let the biggest one stay unregulated on a technicality.

“At least 45 million average monthly users” — the threshold, met. The category was the open question; the size was never in doubt.

Watch three things from here. Whether OpenAI contests the search-engine characterization or simply complies — a fight over category would be a gift to every other assistant. Whether Gemini receives a parallel designation, which would make this a market-wide regime rather than a one-off. And whether the risk assessments, when published, treat web-search synthesis as a distinct systemic risk — because that is the paragraph every other generative company will copy.