The EU classifies ChatGPT as a Very Large Online Search Engine — here's what that means

The European Commission has classified OpenAI's ChatGPT as a Very Large Online Search Engine (VLOSE) under the EU's Digital Services Act, while designating Reddit and Roblox as Very Large Online Platforms (VLOPs). The Commission's official designation list was updated on 31 August 2026, placing all three services under the DSA's most stringent compliance regime (The Verge; EC DSA policy page).

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 threshold triggers obligations that go well beyond baseline DSA requirements: designated services must mitigate risks stemming from the design, functioning, or use of their service, restrict ad targeting to minors and ban targeting based on sexual orientation, religion, ethnicity, or political beliefs, and provide transparency about their recommendation algorithms (The Verge; EC memo on information interference).

ChatGPT, Reddit, and Roblox have until the end of December 2026 to comply with those obligations (The Verge).

Henna Virkkunen, EU executive vice-president for tech sovereignty, security and democracy, said the designations mean the three services "will be held to a higher standard of scrutiny and accountability in the EU" (The Verge).

The VLOSE classification for ChatGPT places it alongside Google Ireland Ltd. and Microsoft Ireland Operations Limited, both already designated as Very Large Online Search Engines on the Commission's updated list (EC VLOP/VLOSE list). The VLOP list now includes Reddit and Roblox among a roster of designated platforms that already covered AliExpress, Amazon, Apple Distribution International, Aylo Freesites, Booking.com, Infinite Styles Services, LinkedIn, Meta Platforms, and NKL Associates (EC VLOP/VLOSE list).

LinkedIn's designation, for context on the user-volume scale involved, was based on 45.2 million logged-in active users and 132.5 million logged-out site visits in the EU (EC VLOP/VLOSE list).

The broader regulatory landscape around AI in the EU has been tightening in parallel. The EU AI Act entered into force on 1 August 2024 and became applicable on 2 August 2026, with exceptions covering prohibited AI practices and AI literacy obligations (EC AI regulatory framework). Separately, on 8 February 2026, the Commission notified Meta of possible interim measures to reverse the exclusion of third-party AI assistants on its platforms (EC announcement).

The ChatGPT designation is the detail that demands the closest attention. The DSA's framework was drafted with conventional search engines in mind: services that crawl, index, and rank web content in response to user queries. ChatGPT does not operate a crawl-and-index pipeline in the manner of Google or Bing. It generates synthesized natural-language responses to prompts, drawing on training data and, in some configurations, live web retrieval. Classifying it as a search engine under the DSA effectively treats an LLM-based conversational interface — a large language model that produces text in response to user input — as functionally equivalent to a keyword-driven index for regulatory purposes.

In this author's view, that framing choice has consequences worth tracking. The DSA's transparency requirements for recommendation algorithms were designed around ranking and personalization systems that shape what users see within a platform or in search results. Applying that lens to a generative model raises questions the DSA's drafters did not squarely address. What does "algorithmic transparency" mean when the algorithm is a transformer producing novel output rather than sorting existing content? How do ad-targeting restrictions map to a service that does not currently serve ads in the conventional sense but may integrate them into generated responses? The designation does not answer these questions, but it places them squarely on the compliance agenda.

The timing compounds the complexity. ChatGPT's December 2026 compliance deadline arrives roughly four months after the AI Act became applicable. OpenAI now faces concurrent obligations under two EU regulatory regimes with different scopes, definitions, and enforcement mechanisms. The DSA governs how the service operates as an intermediary for information access; the AI Act governs how the underlying model is developed and deployed. Where those regimes overlap or conflict is unsettled.

For Reddit and Roblox, the path is more familiar. Both are conventional platforms with user-generated content, recommendation feeds, and advertising models that map cleanly onto the DSA's existing VLOP framework. Their compliance challenge is substantial but not novel: it mirrors what Meta, X, and TikTok have already navigated. Reddit's community-moderation model and Roblox's largely underage user base will each raise specific risk-mitigation questions, but the regulatory structure itself is well-established.

What the designations enable, on a longer view, is regulatory clarity. ChatGPT has been operating in a gray zone, deployed at scale across the EU without a clear classification under the DSA's tiered system. That ambiguity is now resolved. Whether the search-engine category proves adequate for a generative AI service is a question that the next four months of compliance work will begin to answer.