The EU classified ChatGPT as a search engine under its toughest platform law -- and OpenAI's own transparency numbers are what triggered it

The European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act on August 31 -- the first time a generative AI chatbot has received the classification. The designation didn't come from a Commission investigation into how ChatGPT behaves. It came from OpenAI's own transparency disclosure, which put the product's EU reach at 159.1 million average monthly active users over the six months ending March 31, 2026 -- more than triple the 45 million threshold that requires designation under the law.

Reddit and Roblox were designated Very Large Online Platforms in the same batch, at 57.2 million and 46.6 million EU users respectively -- Roblox clearing the bar by just 1.6 million. All three companies self-reported the numbers that triggered their own designation: the DSA already required each of them to publish EU user counts twice a year, and this round simply crossed the line.

What makes a chatbot count as a search engine rather than an ordinary platform comes down to one function, not a product category. The Commission classifies ChatGPT as a hybrid service -- a conversational interface that, when it fetches live results from the open web to answer a query, performs the same retrieval role Google Search or Bing do. That's a capability-based test: it doesn't turn on what OpenAI calls the product, only on whether it searches the web on the user's behalf. (The DSA's toughest tier was written for search engines and social platforms years before generative AI chatbots existed -- ChatGPT is the first case where the law has had to decide whether an AI assistant's web-search feature counts the same way.)

The 45 million figure isn't arbitrary. It's set at roughly 10% of the EU's population, the threshold Article 33 of the DSA uses to define when a service is large enough to be treated as a systemic risk rather than an ordinary business. That's also why the law doesn't much care whether the qualifying product is a social feed, a subreddit, or a chatbot's citation list -- once a service reaches a tenth of the bloc, the Commission treats it as capable of shaping public discourse and safety at societal scale, whatever the company calls it.

"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." — Henna Virkkunen, European Commission executive vice-president for tech sovereignty, security and democracy

The obligations that come with the label are substantial. Designated services must run annual systemic-risk assessments covering illegal content, harm to minors, mental- and physical-wellbeing effects, fundamental rights, electoral integrity and public security; submit to independent external audits; and share data with regulators and vetted outside researchers. Non-compliance can draw fines of up to 6% of a company's global annual turnover -- the same enforcement ceiling that has already produced roughly €870 million in DSA fines across other platforms, including a €550 million penalty against AliExpress.

Exactly when those obligations bite is where the public record gets messy. Every outlet reviewed for this piece agrees the deadline is four months from the August 31 designation -- but the specific calendar date named for it ranges from late November to January 2027 depending on which report you read, and none of them is the Commission's own published notice. Four months from August 31 lands at the end of December, which is the reading this piece trusts, though neither OpenAI nor the Commission has stated the date directly in anything this desk could locate.

The bigger signal here is the precedent, not the paperwork. ChatGPT is the first generative AI product to cross into the DSA's strictest tier, but the test the Commission applied -- does the product search the live web on a user's behalf, and does it reach enough people while doing it -- is written in capability terms, not brand names. Google's Gemini, Anthropic's Claude, and Perplexity all offer some version of live web search inside a conversational interface; any of them crossing 45 million EU monthly users would face the identical designation on the Commission's own stated logic, regardless of how each company markets the feature.

None of this pauses ChatGPT in the EU or forces immediate product changes -- the DSA's toughest tier is about ongoing risk management and outside scrutiny, not a ban. What it does is put a compliance clock, real fines, and an outside auditor's eyes on the product with the largest EU user base of any AI chatbot, at the exact moment the Commission has signaled it's willing to point the same machinery at whichever AI assistant grows into the threshold next.