ChatGPT DSA: AI assistant enters search oversight

Search used to mean a box, a list of links, and a familiar fight over ranking. The European Commission has now looked at ChatGPT, noticed that it can answer prompts and search the web, and put it in the Digital Services Act bucket for very large online search engines. That is the counterintuitive part. The chatbot is being treated less like a private chat window and more like discovery infrastructure. For builders, the useful question is not whether the label feels intuitive. It is what the label makes OpenAI document, audit, and explain in Europe. The DSA does not regulate vibes, despite occasional evidence to the contrary in conference panels.

What the Commission actually labeled ExplainX reported that on Aug 31, 2026,

the European Commission designated ChatGPT as a Very Large Online Search Engine under the DSA, while Reddit and Roblox were designated as Very Large Online Platforms. Gizmodo described the threshold behind those labels as at least 45 million monthly users in the European Union. Winzheng reported a more specific ChatGPT figure, saying the Commission determined that ChatGPT had 159.1 million EU monthly active users. The Commission’s theory, as quoted by EuropeSays, is straightforward: “ChatGPT is an AI system that can engage with and respond to users’ prompts and queries, including by searching the web. Hence, ChatGPT is a hybrid service that qualifies as an online search engine under the DSA.” That sentence is doing the legal work. It is not saying every AI assistant is a search engine. It is saying this assistant, with web search capability and very large EU reach, fits the online search engine definition.

Why a chatbot became search infrastructure Live Index also attributed

the classification to ChatGPT’s ability to respond to prompts and queries while conducting web searches. That matters because the DSA’s largest service categories are not organized around product mascots or user interface metaphors. They are organized around scale and function. If a service helps users discover information at very large scale, the regulator is unlikely to be impressed by the fact that the interface looks conversational. Crypto Briefing reported that OpenAI’s chatbot became the first standalone AI service designated this way. That makes the designation useful for AI product teams even if they are nowhere near the DSA threshold. The lesson is that adding retrieval, web browsing, ranking, or answer synthesis can move a product into a search style compliance conversation. The compliance memo should follow the function, not the marketing category.

What changes in practice Winzheng reported that

the VLOSE designation brings the DSA’s strictest compliance requirements, including systemic risk assessments, independent audits, and data sharing obligations. In plain English, that means OpenAI must be ready to identify risks at scale, show its work to outside reviewers, and support regulator access to relevant data. The details will live in policies, logs, audit trails, and governance processes, not in a cheerful trust center paragraph. ExplainX framed the designation as starting a four month compliance clock, and Superpower Daily used the same four month framing in its coverage. That is a short window if the work has not already begun. For teams that build on top of ChatGPT, the immediate practical question is narrower but still real: what evidence can your vendor provide when your own customers, procurement teams, or regulators ask how search enabled AI outputs are governed in the EU? This is where LinkedIn commentary usually overstates the rule. The DSA designation does not automatically turn every wrapper app into a Very Large Online Search Engine. It does, however, make search style AI governance part of the enterprise checklist: risk documentation, audit cooperation, escalation paths, and clear allocation of responsibility in vendor contracts.

Reddit and Roblox show the label logic Gizmodo reported that

Reddit and Roblox crossed the threshold for designation as Very Large Online Platforms on the same announcement cycle. Their labels are different from ChatGPT’s, but the pattern is the same: once a service is large enough in the EU, the DSA stops treating it as ordinary online plumbing. The category determines the obligations, and the threshold determines when the category becomes painful. EuropeSays likewise reported that Reddit and Roblox were designated as VLOPs while ChatGPT was designated as a VLOSE. That split is the point. The EU did not create an “AI chatbot” bucket for ChatGPT in this move. It used the existing search engine bucket, which should make every assistant builder with retrieval ambitions check how their product is actually used, not how the landing page describes it. The next thing to watch is not a speech about responsible AI. Watch the documents: OpenAI’s DSA disclosures, audit posture, data access handling, and any Commission follow up once the four month clock matures. If you build AI products for European users, the boring takeaway is the useful one: map discovery functions early, keep evidence of risk controls, and do not wait for a designation letter to learn what your system is doing.