The Chatbot Europe Called a Search Engine

Why 'search engine'

The Commission called ChatGPT a hybrid service. It answers prompts and queries, including by searching the web, so it counts as an online search engine for DSA purposes. That move sidesteps a waiting game for the AI Act to invent a chatbot category. The old statute had two drawers, platforms and search. Brussels chose the second drawer and shoved a conversational system into it.

Designation is not a finding of wrongdoing. It is a compliance trigger. Within four months of notification, OpenAI must produce a systemic-risk assessment covering illegal content, minors, mental and physical wellbeing, fundamental rights, elections, and public security. It must take annual independent audits, share data with regulators, and open qualifying research access to vetted outsiders. Fines for later breaches can reach 6 percent of global annual revenue.

Henna Virkkunen, the Commission's executive vice-president for tech sovereignty, framed the decision as size meeting scrutiny. The interesting philosophical claim hides under the bureaucracy. Europe is saying that a system which generates answers at search scale owes the same transparency duties as a system that merely indexes pages. Generation does not buy you a regulatory holiday.

That claim will be tested in the first risk reports due before year-end. A framework written for hosting and ranking now has to describe a model that invents sentences. If the assessments are theater, the designation was theater. If they force OpenAI to document how ChatGPT search steers belief and attention, the label will have done real work.

Sonar has already watched labs measure how models steer beliefs and money. The DSA is attempting something blunter: force the largest conversational search surface in Europe to write that measurement down for the state. Calling ChatGPT a search engine is a category trick. The trick may be the only way an old statute can catch a new machine.