The European Union is having another look at ChatGPT. This time, it’s not about the AI Act — it’s about the Digital Services Act, and the reclassification could mean significantly stricter rules for OpenAI.

According to reports from Handelsblatt cited by DevDiscourse, the EU is examining whether to classify ChatGPT as a “very large search engine” under the DSA. This is a meaningful shift in how the bloc conceptualizes AI chatbots. Rather than treating them as standalone AI products subject to the AI Act’s risk-based framework, the EU is now considering whether platforms like ChatGPT function more like search engines — in which case, they’d face the DSA’s full regulatory apparatus.

What’s the difference? Under the DSA, very large search engines have obligations around content moderation, algorithmic transparency, systemic risk assessments, and data access for researchers. These are the rules designed to force big tech companies to be more accountable for what appears in their results. If ChatGPT is classified this way, OpenAI would have to open up to auditing, provide more data to researchers, and potentially face content restrictions that the AI Act never contemplated.

The timing matters. The AI Act’s first enforcement deadlines kick in over the coming months. The EU has been signaling that it’s taking a hard line on AI governance. Reclassifying ChatGPT as a search engine would be a way to apply pressure without waiting for the AI Act’s full implementation.

This matters beyond OpenAI. If ChatGPT is a search engine, then Claude is a search engine. Gemini is a search engine. Any AI chatbot that people use to find information could face the same classification. The EU would be effectively saying: if your AI system functions as an information intermediary, you’re not an AI product — you’re a platform, and platforms get regulated as platforms.

OpenAI declined to comment, which is unsurprising. What they do next — whether they fight the classification, comply proactively, or try to structure ChatGPT differently to avoid the label — will tell us a lot about how seriously the company takes European regulation.

For the EU, this is a clever regulatory move. They’re not banning AI or even restricting the technology directly. They’re recharacterizing what AI products are, and in doing so, applying rules that already exist. It’s the kind of regulatory arbitrage that European bodies have become相当 good at.

The broader implication is that the EU doesn’t trust AI companies to self-govern even under the AI Act’s lightertouch framework. If ChatGPT becomes a search engine for legal purposes, it signals that the bloc is willing to reinterpret existing law to cover AI systems that don’t fit neatly into established categories.

This is probably just the beginning. Watch for more reclassifications as the EU tries to figure out how existing regulatory frameworks apply to technologies that didn’t exist when those frameworks were written.