On August 31, 2026 the European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act — the European Union’s platform-regulation law — on a declared 159.1 million average monthly active users in the EU. If you saw the headline and want the practical answer: yes, this eventually changes something you can use, but later and indirectly. The designation itself moves no ranking and no citation. What it triggers is a compliance clock, due four months from notification, at the end of which ChatGPT owes the same tier obligations Google Search already carries.
That last part is the half of this story that belongs on a marketing site. The AI-search visibility field currently works from outside measurement — crawl studies, citation sampling, vendor trackers. A regulator has now put ChatGPT in a category where vetted-researcher access to platform data is a legal obligation, not a courtesy. Section 03 maps each obligation to what it could let you see.
- 01ChatGPT is now, legally, a search engine in the EU.The European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act on August 31, 2026, alongside Very Large Online Platform designations for Reddit and Roblox.
- 02The trigger figure is self-declared: 159.1 million.The Commission’s registry lists 159.1 million average monthly active EU users for ChatGPT — declared by OpenAI, not measured by the regulator, against a 45 million designation threshold.
- 03The marketer-relevant duty arrives by end of December 2026.Vetted-researcher data access is due four months from notification — per the Commission, by the end of December 2026. Nothing is disclosed yet.
- 04Non-compliance carries real exposure: up to 6% of turnover.The Digital Services Act caps fines at 6% of annual worldwide turnover for substantive breaches, and reported fines against three designated platforms sum to roughly €870 million — a derived total, not an official Commission figure.
01 — The DesignationWhat the Commission actually did on August 31. #
The action came in the Commission’s own Daily News bulletin of August 31, 2026: ChatGPT was designated a Very Large Online Search Engine — VLOSE, in the Digital Services Act’s vocabulary — while Reddit and Roblox were designated Very Large Online Platforms, or VLOPs. The two categories carry the same heavyweight obligations tier; the difference is what the service is. A VLOP is a very large platform that hosts and recommends content. A VLOSE is a very large search engine — a service whose job is to take a query and return ranked results from across the web. ChatGPT was placed in the second category.
Why a chatbot counts as a search engine is the interesting legal move, and it has to be stated with care, because the underlying designation decision has not been published — the Commission’s own registry entry says the decision is not yet available. The Commission’s reported reasoning, carried in near-identical terms by trade coverage including Search Engine Journal, is that ChatGPT is a hybrid service: an AI system that responds to user prompts and queries, including by searching the web, which qualifies it as an online search engine under the regulation.
Trade coverage consistently describes this as the first time an AI chatbot has received the designation — a framing from the press rather than the Commission’s own words, but one the registry supports: ChatGPT is the only entry in the search-engine tier that is not Google Search or Microsoft Bing.
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. We continue to watch the digital landscape closely and will not hesitate to designate any platform that meets the threshold for enhanced supervision under the Digital Services Act.— Henna Virkkunen, Executive Vice-President, European Commission, August 31, 2026
not yet published. The Commission’s registry lists the August 31 designation with the note that the decision document is not yet available; only the press bulletin and the registry entry exist. Every characterisation of the Commission’s reasoning in this post is therefore reported, not quoted from the decision itself.
02 — The NumbersThe numbers behind the designation. #
The Digital Services Act’s threshold for the very-large tier is 45 million average monthly active users in the EU. Per the Commission’s bulletin, all three services declared figures above it. The specific numbers come from the Commission’s own designation registry, updated August 31, 2026: ChatGPT at 159.1 million, Reddit at 57.2 million, Roblox at 46.6 million. For ChatGPT, the reference period behind the figure is the six months ending March 31, 2026, per OpenAI’s own disclosure as reported by Search Engine Journal.
One caveat carries the article’s quiet irony. These figures are declared by the companies, not measured by the Commission. The numbers that triggered the EU’s strictest transparency regime are, at the point of designation, exactly the kind of self-reported, externally unverifiable claim that the regime’s researcher-access provisions exist to open up.
Declared EU reach · the very-large tier’s newest members in context
Declared average monthly active EU users, per the European Commission’s designation registry, updated August 31, 2026.ChatGPT enters the search-engine tier at a declared 159.1 million EU users — under half of Google Search’s declared 364 million, but more than three and a half times the threshold, and large enough that the regulatory club of search engines in the EU now has an AI assistant as a full member. OpenAI itself, in its statement on the designation, noted that ChatGPT is one of three search services designated as VLOSEs, alongside Google Search and Microsoft Bing.
03 — The PayoffThe duty that matters if you have organic traffic. #
Most of what VLOSE status requires is platform-safety machinery: systemic-risk assessments, annual independent audits, an internal compliance function. One obligation in the Commission’s own plain-language summary of the very-large tier is different in kind for anyone whose business depends on being found: vetted-researcher data access. Article 40 of the Digital Services Act names VLOPs and VLOSEs side by side and requires them to give vetted researchers access to data for research into systemic risks and the effectiveness of mitigation measures, upon a reasoned request from the Digital Services Coordinator of establishment.
Applied to an AI assistant, that duty points at the exact question the AI-visibility field has been guessing at: how does ChatGPT decide which sources to consult, cite and surface? No such data has been opened up — the compliance clock has only started. But the designation puts an AI assistant, for the first time, in the Digital Services Act tier where such access is owed. The table below maps the tier’s obligations, from the Commission’s own summary, to what each one could eventually let a marketer, SEO or researcher see.
| Obligation | What it requires | Why it matters for AI-search visibility |
|---|---|---|
| Vetted-researcher data access (Article 40) | Provide data access to vetted researchers studying systemic risks and mitigation effectiveness, upon a reasoned request from the Digital Services Coordinator of establishment; binds VLOSEs by name | Independent researchers could study citation and sourcing behaviour on platform data rather than sampled output |
| Recommender option not based on profiling | Offer at least one recommender option that does not rely on user profiling | Implies a documented distinction between personalised and non-personalised result selection |
| Systemic risk assessment and mitigation | Identify, assess and mitigate systemic risks the service poses in the EU | A risk assessment for a search service would have to reach how results are produced, not just what content is hosted |
| Public advertisement repository | Maintain a public repository of advertisements shown on the service | If and when ads run in ChatGPT answers in the EU, they would be publicly inspectable |
Obligations in the table are drawn from the Commission’s very-large-tier policy summary; the researcher-access mechanism is from Article 40 of the Digital Services Act. The third column is our interpretation of each duty’s relevance, and it is deliberately conditional — every entry describes what compliance would open up, not what has been disclosed. OpenAI’s own on-record response points the same direction:
ChatGPT search operates as a search service under the DSA, and we are preparing to meet the additional compliance requirements that come with this Very Large Online Search Engine designation.— OpenAI spokesperson, statement to PYMNTS, August 31, 2026
04 — The ClockThe deadline, and what non-compliance costs. #
The obligations bind four months from notification. The Commission’s bulletin puts that at the end of December 2026 — its separate news page phrases the same four-month clock as “by January 2027,” two loose paraphrases of a deadline that lands right at the year boundary. The safer reading, and the one this post uses, is the bulletin’s: end of December 2026.
The enforcement backdrop is not hypothetical. Article 52 of the Digital Services Act caps fines for failing to comply with an obligation at 6% of a provider’s annual worldwide turnover for the preceding financial year, with a separate 1% tier for supplying incorrect, incomplete or misleading information to the regulator. And the Commission has used its powers: across three prior enforcement actions against designated platforms — X at €120 million in December 2025, Temu at €200 million in May 2026, AliExpress at €550 million in July 2026 — fines total roughly €870 million. That is a sum of reported individual fines, not an official running total, but the three components are consistently reported across independent outlets.
Of annual worldwide turnover
Article 52(3) of the Digital Services Act caps fines for failure to comply with an obligation at 6% of the provider’s annual worldwide turnover for the preceding financial year.
The information-offence tier
A separate, lower ceiling applies for supplying incorrect, incomplete or misleading information, failing to reply, or failing to submit to inspection — distinct from the substantive-breach ceiling.
Across three prior fines
X (€120M, December 2025), Temu (€200M, May 2026) and AliExpress (€550M, July 2026) — a derived sum of the three reported fines against designated platforms, not an official EC total.
05 — BoundariesWhat this designation is not. #
Three boundaries keep this story accurate. First, it is not the EU AI Act. The Digital Services Act governs platforms and search engines — systemic risk, content moderation, ranking transparency, researcher access. The AI Act is a different law that governs AI models themselves — risk tiers and transparency obligations for general-purpose AI. OpenAI now carries obligations under both, on separate compliance tracks with separate penalty regimes. Our coverage of who enforces the EU AI Act and with what penalties covers that other track; nothing in this designation flows from it.
Second, it is not a ruling on liability for what AI search says about anyone. That question is being tested elsewhere — a German court ruling on Google’s AI Overviews raised exactly the who-answers-for-the-answer problem, and our analysis of who owns an AI Overview covers it. The Digital Services Act route is different: it regulates the system’s transparency and risk posture, not the correctness of individual answers.
Third, it is not a visibility playbook — yet. We have covered AI search from the measurement side at length, including how AI search traffic and market share have actually grown. This post covers the classification; the measurement posts stand unchanged beside it.
Digital Services Actregulates ChatGPT as a search engine: researcher access, systemic risk, up to 6% of turnover. The
EU AI Actregulates the models behind it: risk tiers and general-purpose AI transparency. Different laws, different clocks, different penalty regimes. This designation lives entirely in the first.
06 — ActionWhat to do with a story whose payoff is four months out. #
Change nothing in your AI-search work today. The designation moves no ranking, no citation and no crawler behaviour. Optimising for AI-assistant visibility is the same problem the day after this designation that it was the day before — which is exactly why the right move now is calendar work, not strategy work.
Watch the compliance window, not the headline. The dates that matter are the end of December 2026, when the obligations bind, and whatever OpenAI publishes to satisfy them — read the primary document, not the coverage of it. The same applies to the first vetted-researcher studies that access under Article 40 would produce.
Treat declared numbers as declared. The 159.1 million figure is OpenAI’s own declaration for the six months ending March 31, 2026. It is the best available figure for ChatGPT’s EU reach and fine to cite — with its provenance attached, the way this post does.
Fold the regulatory lens into visibility work. If AI-assistant visibility matters to your pipeline, what the compliance clock opens up is a scheduled upgrade to how much of that work can rest on evidence rather than inference. Our agentic SEO service already builds AI-search visibility programmes on measured citation behaviour; independent research on platform data would sharpen exactly that input.
07 — ConclusionA category change, with the disclosure still to come. #
Legally a search engine. Practically, a four-month clock.
On August 31, 2026 the European Commission put ChatGPT in the same regulatory category as Google Search, on the reported reasoning that an AI system answering queries — including by searching the web — is a hybrid service that qualifies as an online search engine. The designation rests on a declared 159.1 million monthly EU users, more than three times the 45 million threshold.
For anyone with organic traffic, the designation itself is the less important half of the story. The obligation it triggers that matters most here — vetted-researcher access to platform data — is due by the end of December 2026, and it would be the first regulated route into how an AI assistant picks and ranks its sources. Nothing has been disclosed yet, the legal decision is not even published yet, and every claim about what the disclosures will contain is premature. The accurate posture is the one this post takes: nothing to change today, one clearly dated window to watch, and a penalty regime — up to 6% of worldwide turnover, with roughly €870 million already levied across three prior fines — that makes compliance the expected outcome rather than a hope.