EU Rules ChatGPT a Search Engine on August 31: GEO Gains

News : On 31 August 2026, the European Commission designated ChatGPT a “very large online search engine” under the Digital Services Act (DSA). A very large online search engine (VLOSE) is, under the DSA, a search service exceeding 45 million average monthly EU users, which triggers the strictest tier of obligations. It is the first AI chatbot to receive that status. Reddit and Roblox were designated “very large online platforms” the same day (European Commission, 31 August 2026).

The headline looks like a technicality. A legal label attached to ChatGPT will not move a single position inside a generative answer. What matters is the list of duties that comes with it, and the date those duties bite.

The number that triggered the designation

ChatGPT's search function reached roughly 159.1 million average monthly active recipients in the European Union over the six months ending 31 March 2026. In our assessment the figure is worth checking at source, so we read it in OpenAI's own DSA disclosure rather than in press write-ups. It is not an analyst estimate: OpenAI Ireland Limited published it itself, under the disclosure duty set out in Article 24(2) of the DSA.

The designation threshold is 45 million average monthly EU users. ChatGPT clears it by more than three times. Reddit (around 57.2 million) and Roblox (around 46.6 million) also cross the line, but in the platform category rather than the search engine one.

The legally interesting part fits in one sentence. The Commission treated ChatGPT as a hybrid service: because it answers queries by actively searching the web, it behaves closely enough to a search engine to carry the same status. This is the first time a European regulator has settled a question the industry has left open for two years.

What OpenAI must deliver before January 2027

Designated services get four months from notification to meet the enhanced obligations. The Commission puts the deadline at January 2027. In the Commission's own terms, the regime requires them to:

1. assess and mitigate systemic risks stemming from the service and its algorithmic systems;
2. be audited by an independent auditor at least once a year, and act on the auditor's recommendations;
3. offer a recommender system option that is not based on user profiling;
4. allow vetted researchers to access service data for systemic-risk research;
5. maintain a publicly available advertising repository;
6. share data with the Commission and national authorities.

Skimmed quickly, that reads like more compliance paperwork. Read by someone whose job is to work out why an AI cites one source over another, it reads very differently.

Why GEO gains more than ChatGPT loses

Here is the honest state of the discipline. Everything we know about how ChatGPT picks its sources comes from third-party studies run on samples of a few thousand queries, reproduced from the outside, with no access to the system. We cite them regularly, including the analysis of sources cited by ChatGPT's thinking mode. They are the best data available. They are also outside observation.

Vetted researcher access changes the nature of that evidence base. For the first time, academic work will be able to rest on data supplied by the service rather than reconstructed by sampling. It is the same movement we saw on the UK side when the CMA imposed ranking transparency duties on Google: a regulator opens a box the market could not open on its own.

Reddit's designation matters just as much. Reddit is one of the most cited domains in generative answers, and its citation share swings hard: we documented an 86% collapse in its ChatGPT citations over four days. Having a link that central to AI visibility placed under annual audit and researcher access adds a source of truth for everyone.

What changes for you, and what does not

Start with what does not move. Your presence in ChatGPT answers is exactly the same on 1 September as it was on 30 August. No algorithm change was announced, and the designation mandates none, so we don't expect any ranking movement before the rules bite. Anyone selling you a “DSA compliance package for your AI visibility” this week is selling you air.

What does move is slower and more useful. From January 2027, three things become enforceable: documentation of the parameters driving recommendations, an annual audit whose findings bind OpenAI, and a legal channel for researchers to ask precise questions against real data. GEO moves from guesswork to evidence, with a date attached.

Until then, three actions make sense. Keep producing structured, sourced, attributable content, because that is what models cite today and nothing in this decision changes it. Measure your AI citations now, so you hold a baseline from before the rules take effect. And keep your pages reachable by generative crawlers, a topic we covered in depth through the lens of the AI content labelling duty in the AI Act.

The limits of what we know

The Commission's release describes a regime of obligations. It does not say how those obligations will apply in practice to a generative engine, and the question is far from trivial. Documenting the “main parameters” of a conventional recommender system is a known exercise. Doing it for a language model that composes an answer from sources retrieved on the fly has no precedent.

Second caveat: nothing in the release guarantees that transparency will cover citation selection, which is the part that concerns us. The duties are built around systemic risk, illegal content and the protection of minors, not fairness between publishers. Third caveat: January 2027 is a compliance deadline, not a publication date. What the public gets to see, and when, is still undetermined.

The Cicero take

The designation has no effect on your visibility this week, and a considerable effect on the credibility of GEO eighteen months out. A discipline that runs on samples and reverse engineering has just been handed, through regulation, the promise of a factual base. We would far rather work with that.

Frequently asked questions

Alexis Dollé

CEO & Founder

Growth and SEO & GEO content strategist, I founded Cicéro to help businesses build lasting organic visibility : on Google and in AI-generated answers alike. Every piece of content we produce is designed to convert, not just to exist.

LinkedIn