The EU has a funny way of making things sound boring while completely shifting the ground beneath an industry. In that tradition, there’s a story making headlines (well, Handelsblatt headlines) that OpenAI’s ChatGPT might soon be classified as a “very large search engine” under the EU’s Digital Services Act.

Sounds technical. Actually matters quite a lot.

What VLSE actually means

Under the DSA, “very large search engines” - think Google Search, Bing, and their ilk - face a significantly higher regulatory burden than your average platform. We’re talking mandatory risk assessments, transparency requirements on algorithmic ranking, and obligations to mitigate systemic risks. The logic is that when a platform serves hundreds of millions of users and shapes what information they see, it carries outsized responsibility for that influence.

Now imagine applying that framework to ChatGPT. Not because it indexes the web - it doesn’t, in the traditional search engine sense - but because it processes queries from roughly 200 million monthly users worldwide and generates responses that shape how people understand the world.

Why this matters now

The EU has been relatively quiet on how exactly the AI Act and DSA interact with generic AI models like ChatGPT. The AI Act targets high-risk systems. The DSA targets platforms with “large online presence.” The gap has been: what about general-purpose AI products that behave like search engines but aren’t technically searching anything?

This classification would close that gap. If ChatGPT is a search engine, OpenAI suddenly faces obligations around how it ranks and presents information, what transparency it provides about its algorithmic choices, and potentially requirements around content moderation and misinformation that apply to VLSEs but not to regular AI developers.

The broader picture

This is the EU doing what the EU does best: creating regulatory frameworks that define terms before companies define them themselves. WhetherChatGPT is technically a search engine or not misses the point. The functional question is: does a product that responds to 200 million queries a month with AI-generated answers carry similar societal responsibilities to a search engine?

The EU thinks yes. OpenAI hasn’t commented. The smart money says this classification fight is just beginning.