The EU just handed OpenAI its first real content moderation headache — the kind Facebook's been dealing with for a decade.
The Summary
- ChatGPT is now classified as a Very Large Online Search Engine under the EU's Digital Services Act, joining Reddit and Roblox as newly regulated platforms after crossing 45 million monthly EU users.
- OpenAI must now mitigate risks to minors, mental health, and illegal content spread, the same compliance burdens that cost Meta and Google billions in legal and engineering resources.
- The DSA bans targeting ads based on sexual orientation, religion, ethnicity, or political beliefs, and prohibits targeting minors entirely.
The Signal
This marks a turning point for AI regulation. OpenAI has operated in a gray zone where ChatGPT wasn't quite a search engine, wasn't quite a social platform, wasn't quite a publisher. The EU's Digital Services Act just collapsed that gray zone. At 45 million monthly EU users, ChatGPT triggered the threshold that pulls it into the same regulatory framework as Google Search and Facebook.
The practical implications are massive. OpenAI now needs content moderation teams, appeals processes, transparency reports, and risk assessment frameworks. The mental health and minors provisions are particularly thorny. How do you prove a conversational AI isn't harming a teenager's mental health? What counts as "illegal content" when the model generates text that never existed before?
"OpenAI will soon be held accountable for mitigating risks related to ChatGPT's impact on minors, user mental health, and the spread of illegal content."
The advertising restrictions hit differently for AI than for traditional platforms. ChatGPT doesn't run banner ads, but it does have a business model problem. If OpenAI can't personalize based on the rich behavioral and preference data ChatGPT collects, the path to an ad-supported tier just got harder. Reddit and Roblox face the same constraint, but they've been building moderation infrastructure for years. OpenAI is starting from scratch.
Here's what most coverage misses: this designation isn't about what ChatGPT does today. It's about what the EU expects it to become. Classifying it as a search engine signals that Brussels sees conversational AI as a primary information gateway, not a productivity tool. That framing matters. It means ChatGPT gets regulated like infrastructure, not software.
The timing is telling. The 45 million user threshold suggests ChatGPT's EU growth has been climbing faster than OpenAI publicly acknowledges. The company doesn't break out regional user numbers, but crossing this line puts them in the same league as Reddit and Roblox, platforms with deep engagement hooks and network effects. For an AI interface, that's a different scale than expected.
Key compliance burdens now facing OpenAI:
- Quarterly risk assessments for illegal content, mental health harms, and minor safety
- Independent audits of algorithmic decision-making and content moderation
- Public transparency reports on takedown requests and enforcement actions
- User appeals systems for content or account restrictions
The Implication
Watch how OpenAI responds in the next 90 days. If they spin up a Brussels policy team and start hiring content moderators, they're treating this as the new normal. If they push back or seek exemptions, they're betting that AI exceptionalism still has legs in regulatory circles.
For other AI companies, this is the template. Once your chatbot crosses 45 million monthly users in the EU, you're not a tool anymore. You're a platform. That means legal teams, compliance officers, and the slow bureaucratic drag that comes with every feature launch. The scrappy "move fast" era of AI development just hit a European speed bump.