The European Commission has decided to classify ChatGPT as a platform with the maximum level of regulatory control under the Digital Services Act (DSA). This means OpenAI is now obliged to comply with the strictest requirements for risk management, transparency, and auditing. Notably, Claude from Anthropic did not make it onto this list — and this is not a coincidence, but a natural outcome of formal criteria.
Why the classification depends on numbers
At the core of the decision lies not an assessment of potential danger or the ethics of AI, but solely the scale of the audience. The threshold for "very large platform" status is an average of 45 million users per month in the European Union. Companies provide the data themselves, and Brussels issues its verdict based on that.
OpenAI stated that ChatGPT's search feature is used by 159.1 million people — roughly 3.5 times above the established minimum. For comparison, Reddit reported 57.2 million, while Roblox barely crossed the line, showing a result just 1.6 million above the threshold. As for Claude, its declared audience in Europe turned out to be significantly below 45 million.
This rule works both ways. In May 2025, the Commission stripped the adult service Stripchat of its "very large platform" status — its audience had fallen below the critical mark.
"We will not hesitate to add to the list any platform that reaches the threshold for enhanced oversight under the Digital Services Act," said Henna Virkkunen, Executive Vice-President of the European Commission for Technological Sovereignty.
Why Claude from Anthropic remained below the 45 million threshold
Anthropic is also required to disclose data on its audience at least twice a year. However, during the reporting period ending October 31, 2025, the number of Claude users in Europe did not reach the threshold. The company does not disclose the exact figure, and this data is already ten months outdated. It is time for Anthropic to publish fresh statistics — especially given that it has announced plans to go public on the stock exchange against this backdrop.
All three services that made the list are required to begin complying with the new rules within four months, i.e., by January 2027. They will need to conduct assessments of systemic risks related to illegal content, disinformation, mental health, and elections, as well as audit their own algorithms.
The Commission is now applying the law seriously, not just for show. In December 2025, it fined X (formerly Twitter) 120 million euros — the first instance of punishment for violating the DSA. The law itself provides for fines of up to 6% of a company's global turnover.
Claude is currently off the list only because of its scale. Anthropic has already faced a lawsuit over training data, and the company's next report in Europe will show whether it falls under the strictest section of the DSA.
My take: this is an important signal for the entire AI industry. The formal criterion of 45 million users is a rather crude tool that does not account for a model's real impact. However, it forces companies to compete more actively for audiences while simultaneously preparing for stricter regulation. For investors, this means that scaling AI products in the EU will be directly tied to growing regulatory burdens.