EU Puts ChatGPT, Reddit, Roblox Under DSA

EU’s Strictest Rules Reach ChatGPT

The European Commission moved ChatGPT, Reddit and Roblox into the top tier of the Digital Services Act on Monday. All three declared that they reach at least 45 million monthly users in the European Union. That figure is the trigger. Each service now has four months to meet a longer list of duties.

Key Takeaways

ChatGPT counts as a very large online search engine, while Reddit and Roblox count as very large online platforms.
All three services declared at least 45 million average monthly users across the EU.
The compliance clock runs for four months from notification, landing at the start of 2027.

What the designation demands

Services in this tier must assess systemic risks and then reduce them. Illegal content spreading through the platform is the obvious one. Harm to minors comes next, along with damage to the physical and mental wellbeing of users.

The duties run further than content. Companies at this level have to limit risks to fundamental rights, to electoral processes and to public security. The Commission supervises that work directly and gains investigative powers to examine the functionality behind each service.

Why ChatGPT counts as a search engine

OpenAI’s assistant answers questions by drawing on the open web, so Brussels files it as a search engine rather than a platform. Reddit and Roblox fall on the other side of the line. Both let users push third-party content to a public audience, which is the textbook definition of an online platform under the law. Roblox carries extra weight in the minors category, because a large share of its players are children.

A first for general-purpose AI

The decision stretches a regime built for social networks and marketplaces to cover a general-purpose AI assistant. The Commission’s formal notice published on August 31 hands OpenAI the same obligations already carried by the largest social platforms, including annual risk reporting and independent audits. Nothing in the text is written for chatbots specifically, which leaves both sides working out how a conversational answer engine documents its risk profile.

Europe keeps arriving first with rules for AI products, and companies increasingly build compliance features for the EU and then ship them everywhere. Reporting on how the decision landed across the bloc noted that the three newcomers join a group of platforms already answering to Brussels every year. For Roblox, the practical work centers on child safety systems. For Reddit, it centers on moderation at volume. For OpenAI, almost everything is new.

A pattern in Brussels

Enforcement has been building for two years. TikTok already faces EU charges over ad transparency failures. Under a separate law, regulators want Google to share search data with rivals. Italy has fined an AI chatbot service over weak age checks. ChatGPT enters that queue with the biggest audience of the three, and its free tier keeps adding features that pull in more users every month.

Fines under the law can reach six percent of global annual turnover, though the Commission has to open a formal case first.

Anthropic Resumes External Cybersecurity Testing After Claude Breach Incidents
Open-Weight AI Platforms Become Silicon Valley’s Top Acquisition Targets
Chinese Automakers Pour Billions Into Humanoid Robot Ventures
OpenAI to Cut Off Cursor’s Model Access on November 12 After SpaceX Deal
Anthropic Opens Research Preview of a Standard Letting AI Agents Run Lab Hardware