Europe Decides ChatGPT Is a Search Engine, and Rules Follow

Brussels has settled a question that lawyers have been circling for a year: in the eyes of European law, ChatGPT is a search engine. On August 31 the European Commission designated OpenAI's chatbot a "very large online search engine" under the Digital Services Act, the first time an AI chatbot has been brought inside the bloc's flagship online-safety regime. The same day, Reddit and Roblox were named very large online platforms under the same law, according to Euronews.

The designation turns on size and function. A service crosses the threshold once it has at least 45 million monthly users in the European Union, and OpenAI reported roughly 159 million average monthly recipients for ChatGPT search. The Commission described the product as a "hybrid service" that behaves like a search engine when it answers a query by looking things up on the web, which is enough to qualify. Whether a chatbot is really a search engine has been debated by academics and regulators for months, and Europe has now given its answer in the way that counts.

What the label brings

Being designated is not a fine or a ban. It is a set of standing obligations, and they are demanding. Within roughly four months, ChatGPT will have to identify and reduce "systemic risks" flowing from its service and its algorithms, submit to an independent audit at least once a year, and share data with the Commission and national regulators so outsiders can check its work. For a system that generates answers rather than returning a list of links, spelling out how to assess and mitigate risk is going to be genuinely hard, and the answers OpenAI gives will set expectations for every rival that hits the same user numbers.

This is the second heavy piece of European enforcement to land on the major labs in a matter of weeks. We recently covered the first formal demands issued under the separate AI Act to OpenAI, Anthropic and Google. The DSA designation comes from a different law with a different logic, aimed at platforms and the risks they spread rather than at the models themselves. Taken together, they show a regulator applying two large rulebooks to the same handful of companies at once.

The precedent problem

The reasoning is what other chatbot makers will study. If answering a prompt by searching the web makes a product a search engine, the line will not stop at ChatGPT. Google's Gemini, Anthropic's Claude and a dozen assistants that fetch live results all do the same thing, and most are climbing toward the 45 million mark in Europe. OpenAI drew the designation first because it is the biggest. It is unlikely to be the last, and the compliance machinery it builds over the next four months may end up being the template the rest are measured against.