Building on our coverage from 10 days ago, the EU has moved to hold ChatGPT, Reddit and Roblox to a higher standard under the Digital Services Act.
The Commission formally designated ChatGPT as a Very Large Online Search Engine, or VLOSE, the first generative AI chatbot to receive that classification. That label obliges the company to perform systemic risk assessments, be more transparent about how the service works, and cooperate with independent researchers and regulators. EU authorities can demand risk audits, require product changes, and impose substantial fines for non-compliance; reports say OpenAI has roughly four months to meet the new rules.
Why this matters: regulators are no longer treating advanced chatbots as experiments. They are treating them as services whose risks must be measured, published and mitigated. For businesses and users that means big providers will have to show their safety work and accept external scrutiny, not just announce updates.
How it works, in one image: imagine a vehicle inspection for online services. Being labeled VLOSE gives regulators the right to ask to see the safety tests, to let vetted researchers probe the system, and to require fixes if harms are found. The technical meaning of the label is less important than the practical powers it hands to authorities.
What changes now: researchers may win better access to evidence about these systems, and regulators have a clear legal route to force changes. That could shift where responsibility sits: from individual users or buyers to the companies that run the services. It does not mean anyone can run ChatGPT locally; operating systems at this scale still needs major infrastructure.
The real test is ahead: will EU enforcement translate into meaningful changes to how these systems behave, or will compliance be limited to paperwork? We'll be watching which way it goes.