News with an accent.

The curved glass-and-steel facade of the Berlaymont building, European Commission headquarters in Brussels, with a large vertical banner on its tower.
El edificio Berlaymont, sede de la Comisión Europea en Bruselas (imagen de archivo).Wikimedia Commons / Trougnouf (Benoit Brummer), CC BY 4.0 · Imagen de archivo
Artificial intelligence

EU asks OpenAI and Anthropic for information under AI Act

The AI Office sent requests after two summer security incidents; no violations have been found and experts dispute whether the law covers testing.

The European Commission's AI Office has started applying the AI Act to OpenAI and Anthropic with formal requests for information, according to Euronews. Henna Virkkunen, the Commission's executive vice-president for tech sovereignty, announced it on Oct. 14 at the International AI Summit in Dublin. The requests are not a penalty: so far there are no violations found and no fines.

The Office is asking for data on safety, external evaluation and monitoring of models in use. The requests respond to two incidents this summer. In one, an OpenAI agent broke out of its test environment and breached Hugging Face, a platform for open models. In the other, a recent Anthropic model gained "unauthorised access" to three outside organizations. The companies supplied the details of both incidents and they have not been independently verified.

Euronews reports that the Office judged the documents the two companies had handed over voluntarily to be insufficient. Another outlet, Borncity, adds a fourth area, training data, which Euronews does not mention; the Commission has not confirmed the difference. No Commission statement was found either: Virkkunen's remarks are known through the press.

The legal dispute

The AI Act has an exemption for research and development. Lawyer Barry Scannell, a partner at William Fry, told Euronews the law does not apply to development and testing activity. An unnamed Commission official said the law stays out of the picture while a model is being tested, but comes in if the model escapes its environment and affects the EU market. Hugging Face has customers worldwide, so the OpenAI case would have touched European users.

The Office has had enforcement powers since Aug. 2, 2026. This is the first time it has used them on two of the largest labs, and what it decides about internal testing will set a standard for any model reaching the European market, including those used in Mexico through the same apps.

It remains to be seen how OpenAI and Anthropic will respond, and by when. According to Borncity, that is the next step in the process.

Keep reading

El Mediático counts visits without cookies, with its own measurement and with Google Analytics, which receives basic data about your visit. Only if you accept does Google Analytics also store cookies and, where advertising is active, show personalized ads. You can accept all, reject all, or choose what to allow. The cookie policy is available in Spanish only. More about cookies (Spanish)