In the second week of June, the deadline for enforcement of European AI law requirements stood at 55 days, a countdown followed closely by providers of general-purpose models.
The content of the requirement is what makes it uncomfortable for the sector: documenting information about the model, providing it to authorities and to downstream builders, maintaining a copyright policy, and publishing a sufficiently detailed summary of the content used in training.
The obligation to publish a training-material summary converts into a legal duty what has always been a commercial decision. Companies avoided detailing data provenance precisely because it opens the door to copyright disputes.
The European calendar ran in parallel with court decisions pointing the same way, with a German court later holding that training on protected repertoire requires a licence, reaching both the training and the similarity of what the model generates.
For anyone building on someone else's model, the rule creates a new information channel: the supplier must pass compliance data down the chain, which helps anyone who has to answer their own customer.