European Commission Executive Vice President Henna Virkkunen revealed that the AI Office has sent Requests for Information (RFIs) to several general-purpose AI model providers as a first step in enforcing the AI Act.
PolicySecurityEuropean Commission
European Commission Requests Information from Model Providers Under the AI Act
This article is a translation. Read the Japanese original
It has been reported that the targets include major laboratories such as OpenAI, Anthropic, and Google.
The requests concern model security, independent external evaluations, and post-market monitoring.
Obligations for general-purpose AI models under the AI Act came into force on 2026-08-02. The European Commission is exercising this legal authority within four weeks of enforcement.
In the event of false or incomplete responses, fines of up to €15 million or 3% of total worldwide annual turnover, whichever is higher, may be imposed. Furthermore, in serious cases, corrective measures such as restrictions on the model's availability in the European market may be required.
This move follows a series of incidents regarding model safety that occurred over the summer. Reports have included OpenAI-related technology intruding into Hugging Face, as well as breaches of external systems by models from Anthropic and Meta.
The European Commission has acknowledged that it is conducting individual consultations with OpenAI and Anthropic regarding these incidents.
Source: The EU has begun enforcing the AI Act: first RFIs to model providers (Hacker News Frontpage, 2026-08-31)