In today’s episode of “We built a general-purpose text machine and accidentally reinvented the internet again”: the EU may start treating ChatGPT’s search feature like a Very Large Online Search Engine.
The European Commission is assessing whether ChatGPT’s search feature should be treated like a “very large online search engine” under the EU’s Digital Services Act — which is bureaucrat-speak for: welcome to the grown-up table, bring audit logs.
What’s actually happening
Under the Digital Services Act (DSA), services that cross 45 million monthly active users in the EU can be designated as a Very Large Online Platform (VLOP) or a Very Large Online Search Engine (VLOSE).
Computing reports the European Commission is assessing whether ChatGPT should be designated as a VLOSE — a category that comes with extra risk-management and transparency obligations.
Why “ChatGPT search” matters (and why the number is the point)
The hook here isn’t philosophical. It’s arithmetic. Both Computing and PYMNTS/CPI say OpenAI published EU user figures for ChatGPT’s search feature above the threshold. PYMNTS/CPI (citing Reuters) reports an average of about 120.4 million monthly active users in the EU over a six-month period ending September 2025.
Once you’re in that user-count weight class, the law stops treating you like “a tool” and starts treating you like “an infrastructure problem”. Congratulations: you’ve achieved systemic relevance. Your participation is becoming increasingly mandatory.
What a VLOSE designation would force OpenAI to do
According to Computing and VinciWorks, a designation would typically mean:
- More formal risk assessments and mitigation plans (systemic risks, not just “we promise we tried”).
- More transparency about policies, moderation choices, and — increasingly relevant — advertising and recommendation dynamics.
- Independent audits and more structured engagement with regulators.
- A short compliance fuse: Computing notes companies generally have four months after designation to comply.
Computing also notes that DSA supervision can include a fee tied to global income (reported as up to 0.05% of global annual net income), plus meaningful penalties for breaches.
The real question: what is ChatGPT, legally?
VinciWorks frames the core tension: the DSA was written for platforms that host/distribute content and search engines that index/retrieve it. ChatGPT blurs the categories by doing retrieval, generation, and interaction all at once. Regulators are trying to pin a moving jellyfish to a taxonomy chart.
And of course, the Commission spokesperson line (reported in multiple places) is that these decisions are made “case-by-case.” That’s regulator for: please stop inventing new product categories faster than we can name them.
What to Watch
- Whether the designation targets “ChatGPT search” specifically (feature-scoped oversight) or the broader ChatGPT product.
- What “transparency” means for an LLM interface: ads, ranking, citations, and whatever we’re calling “answers” this week.
- The precedent effect: once one general-purpose AI product is treated as platform-scale infrastructure, the line moves for everyone else.
The Singularity Soup Take
This is the Brussels Effect in its natural habitat: the moment a tool gets big enough, Europe stops asking whether it’s “a product” and starts asking whether it’s “a public risk surface”. If you wanted the power of a search engine without becoming a search engine, I recommend building smaller things. Or lying about your user numbers. (Do not do that.)
Sources
Computing.co.uk — "EU set to classify ChatGPT under strict online platform rules"
VinciWorks — "EU moves to rein in ChatGPT"
PYMNTS/CPI — "European Commission Reviews Whether ChatGPT Falls Under EU Digital Services Act Rules"