The European Commission has taken a decisive step, officially classifying OpenAI's ChatGPT under the category of platforms with the highest level of regulatory oversight within the Digital Services Act (DSA). This automatically imposes a number of strict obligations on the company related to transparency, risk management, and content moderation. Notably, its main competitor — Claude from Anthropic — did not make it onto this "elite" list. This is not a coincidence, but rather a consequence of a formal yet extremely important criterion.
Why Audience Size Decides Everything
The key parameter for falling under enhanced oversight is not an assessment of the technology's potential danger, but simple mathematics. The threshold is set at 45 million average monthly users in European Union countries. Platforms that surpass this mark are automatically recognized as "very large" and are required to comply with elevated DSA standards.
Here are the figures that led to this decision. OpenAI stated that ChatGPT's search function is used by 159.1 million people in the EU — more than 3.5 times the established limit. The social platform Reddit reported 57.2 million users, while the gaming ecosystem Roblox only slightly exceeded the minimum, showing an audience 1.6 million above the threshold.
The mechanism works in both directions. In May 2025, the Commission stripped the adult service Stripchat of its "very large" platform status, as its audience fell below the critical mark. This demonstrates that the regulator approaches the matter dynamically, rather than simply "attaching labels" permanently.
Claude Remains in the Shadows, But Not for Long
Anthropic, as required by the DSA, publishes data on its European audience twice a year. The latest report, covering the period up to October 31, 2025, showed that Claude does not reach the coveted 45 million. The company, however, does not disclose the exact figure, and this data is already ten months outdated. Given Claude's rapid growth in popularity and Anthropic's recent statements about preparing for an initial public offering, it can be expected that the next report will dramatically change the situation.
Meanwhile, the countdown has begun for OpenAI, Reddit, and Roblox. They have four months — until January 2027 — to conduct a comprehensive assessment of systemic risks related to illegal content, disinformation, users' psychological health, and the protection of minors. This is not a formality: the Commission has already proven the seriousness of its intentions by fining X (formerly Twitter) €120 million in December 2025 — the first fine in the history of DSA enforcement. Potential liability for violators reaches 6% of the company's global turnover.
My view: the current classification is merely a matter of time. The formal criterion based on user numbers does not reflect the technology's real impact, and as soon as Anthropic crosses the threshold — which, judging by growth rates, will happen within the next reporting periods — Claude will come under the same close scrutiny. The industry should prepare for a new era where regulatory audits become an integral part of the lifecycle of any significant AI product.