The European Commission is analyzing responses from more than 30 artificial intelligence companies consulted about model safety, transparency and copyright. The information was confirmed on Friday (9) by Henna Virkkunen, executive vice president responsible for digital affairs, in an interview with Reuters. The official also revealed that Chinese startups received requests, expanding the reach of European oversight beyond the leading Western developers.
The questionnaires were sent at the end of August and initially requested information on the measures adopted by the companies to ensure the safety of their systems. Later, the Commission added questions related to transparency and compliance with copyright obligations.
Virkkunen stated that regulators are evaluating the responses and may consider recommendations from a scientific panel made up of 60 AI experts affiliated with universities and research institutions.
Chinese companies were consulted, but the Commission did not disclose the names
The inclusion of Chinese startups was confirmed directly by Virkkunen, who highlighted the presence of advanced models developed both in the United States and China. According to the official, these systems present new risks that need to be monitored by regulators.
The Commission, however, did not identify any of the companies consulted, whether American, European or Chinese. It also did not say how many startups from China received the questionnaires.
The procedure is part of the oversight provided for in the AI Act, European legislation that establishes obligations for safety, transparency and monitoring of artificial intelligence systems according to their characteristics and risks.
The current stage consists of analyzing the requested information, not formal investigations or sanction proceedings announced against these companies.
The regulation provides for fines that can reach 7% of annual global revenue in certain infringements, especially prohibited AI practices. This limit, however, does not automatically apply to the questionnaires sent by the Commission, and other categories of violation have different penalties.
Virkkunen also rejected criticism that the legislation was outdated in light of recent advances in technology. According to her, the regulation already includes continuous risk assessments, participation of external experts and monitoring of the models.
The Commission did not announce a deadline to complete the analysis of the responses or say whether the questionnaires will result in additional proceedings.
