Europe Just Classified ChatGPT as a Search Engine. The Audit Starts in January.
· CX Pulse
The Commission designated ChatGPT a Very Large Online Search Engine on 31 August, and the obligations that follow have a January 2027 deadline.
Europe Just Classified ChatGPT as a Search Engine. The Audit Starts in January.
On 31 August the European Commission designated ChatGPT a Very Large Online Search Engine under the Digital Services Act. Reddit and Roblox were designated Very Large Online Platforms in the same decision. Most of the coverage has flattened that distinction, and it's the part that carries the consequences.
The Commission's reasoning is a single sentence in the announcement. "ChatGPT is an Artificial Intelligence (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." The designation turns on what the service does rather than what category of technology it belongs to. Any assistant that answers questions at scale can meet that test.
Nobody at the Commission went and measured the traffic. The announcement says these services "declared that they reach at least 45 million average monthly users in the EU and thus meet the threshold for designation." OpenAI's own filing puts ChatGPT search at roughly 159 million average monthly recipients in the EU across the six months to 31 March 2026. The company reported itself into the strictest tier of European platform law.
What the designation actually produces
Four months, so by January 2027. In that window the service has to identify, assess and mitigate systemic risks arising from its service and its algorithmic systems. It has to stand up an internal compliance function. It has to be audited by an independent auditor at least once a year and act on what the auditor finds. It has to share data with the Commission and national authorities, and open a channel for vetted researchers to study systemic risks. It has to offer a recommender option not based on profiling, keep a publicly available repository of advertisements, and report transparently on advertising, recommender systems and content moderation decisions.
The Commission also picks up investigative powers "to assess the functionalities behind these services and, where applicable, any related system." Supervision runs jointly with Coimisiún na Meán, Ireland's Digital Services Coordinator, since that's where OpenAI's EU establishment sits.
Why this reaches customer experience
A growing share of the questions people ask about your business never arrive at your business. They go to an assistant, it answers, and that answer is the entire interaction. There has been nothing external to inspect: no ranking documentation, no index, no published account of how one vendor gets named instead of another. We made that point on 20 August about ChatGPT's unpaid answers and it still holds. What changed yesterday is that the documentation now has a deadline attached to it.
If you run CX at scale
Someone in your company owns Search Console. Someone owns brand monitoring, the review-response workflow, the takedown path for a damaging search result. Nobody owns the assistant layer, because there's been nothing to own.
Three things turn into real inputs. Transparency reporting on recommender systems and moderation is the first outside account of how these answers get ranked, which is what your discovery function has been guessing at. The public advertisement repository is a queryable record of who pays to appear alongside answers in your category. The researcher access channel makes independent studies of how assistants recommend vendors possible, and findings about your category will be published whether or not anyone on your team is reading them.
The question for the next planning cycle is who receives all of it. Discovery and SEO is the obvious home. Legal and compliance has a claim, since the risk assessment and the audit are regulatory documents. Brand has a claim, since the ad repository is a competitor-monitoring surface. Name one owner now instead of finding out in February that three teams each assumed it belonged to another one.
If you run a smaller operation
The same shift arrives with the opposite problem. A large company has controls that may not survive the change. A small one never built them. No SEO function, no monitoring stack, no legal team to send a letter when a machine describes the business wrong.
The advertisement repository is the piece worth caring about here, because it's public and it's free. Once it exists you can look up who is buying placement against questions in your category. For a business that has never run a competitive research project, that is one bookmark and twenty minutes.
Two things to hold lightly
This is EU scope. A company with no European customers benefits only through whatever OpenAI chooses to publish everywhere. And four months is a compliance deadline, not the date useful documents show up in public; reporting cycles run slower than announcements suggest.
The direction is still set. The Commission has now designated 28 very large platforms and search engines, and the test it applied here is about function rather than technology. The layer sitting between your customer and your business is going to be documented. Worth deciding now who reads it.