The European Commission on Monday designated ChatGPT as a Very Large Online Search Engine and Reddit and Roblox as Very Large Online Platforms under the Digital Services Act, the first time a generative-AI chatbot has been folded into the law's strictest tier. [1][2] All three services declared at least 45 million average monthly users in the EU, the threshold that triggers the extra rules. [1][2] They have four months — until January 2027 — to comply. [2]

Reuters treated the move as a compliance brief. The Commission said the three will have to assess and mitigate risks of illegal content, negative effects on minors and on users' physical and mental wellbeing, and risks to fundamental rights, electoral processes and public security. [1] Brussels also gains investigative powers to look at the functions behind the services. [1] The wire's most important sentence is the classification itself: "AI-system ChatGPT qualifies as an online search engine under the DSA." [1] Reddit and Roblox qualify as platforms because they let users disseminate third-party content to the public. [1] A chatbot that answers prompts is being regulated as if it were an index.

That is the first. The largest social platforms already live in this tier. A model that generates answers does not. Once designated, a VLOP or VLOSE must identify systemic risks linked to its service and algorithms, put mitigations in place, run an internal compliance function, submit to an independent annual audit, share data with the Commission and national authorities, and give vetted researchers access when the work concerns those risks. [3] Large services must also offer a recommender option that is not based on user profiling and keep a public ad repository. [3] Mitigations can mean changing a service's design, altering a recommender, or adding staff to find the risks the assessment names. [3] Platforms must update user-number figures at least every six months. The designation can be revoked if the service stays under 45 million monthly users for a full year. [3]

The search-engine label is not a footnote. Reuters said the "AI-system ChatGPT qualifies as an online search engine under the DSA," while Reddit, a discussion-based platform, and Roblox, a gaming platform, qualify because they let users disseminate third-party content. [1] Folding a chatbot into the search bucket puts OpenAI under the same extra-tier duties as an index, including the audit and the researcher-access rule, without first deciding that ChatGPT is a social network. That choice will shape every later fight over whether an answer is content, a search result, or both.

X read the same facts as overreach. An MTS post Monday evening quoted Theo Jaffee: OpenAI will have to publish reports on how the platform "harms, quote, civic discourse, according to EU bureaucrats, and then do mandatory mitigation measures, meaning fix these things in ways that the EU bureaucrats tell them to." That is a real argument about who defines harm. It is also not the news. The news is that no chatbot has been told to do this before. The Commission's own list names illegal content, fundamental rights, electoral processes, public security, minors and wellbeing — not "civic discourse" as a separate heading. [2] Jaffee's phrase is the X compression of that list. Applying the list to a search-classified AI system is the novelty Reuters underplayed and Jaffee over-personalized.

The ordinary DSA duties still apply on the way up: a point of contact, criminal-offense reporting, user-friendly terms, and transparency on ads, recommenders and moderation decisions. [3] The extra tier is the risk assessment and the audit. The Commission fined AliExpress €550 million in July for failing to assess and mitigate risks tied to illegal, unsafe or counterfeit products — a reminder that the extra tier is not hortatory. [3] Four months is a short build for annual systemic-risk reports, an independent auditor and researcher access. ChatGPT, Reddit and Roblox now have until winter to show they can do the paperwork. Whether OpenAI contests the search-engine label or complies inside the four-month window is the next record. The first designation of a chatbot as a systemic-risk search engine is already filed.

-- DAVID CHEN, Beijing