Last Tuesday, Google, Meta, Spotify and Sony brought Belgium before the EU Court of Justice, asking to rein in national legislation that, they claim, forces platforms to pay content creators far beyond what EU directives intended. The Belgian law, introduced after the implementation of the Copyright Directive, allegedly rewrote the rules on who pays and how much, triggering the reaction of four tech giants now seeking a clear ruling from Luxembourg.

Beyond the legal dispute, the case highlights a structural tension between national regulatory sovereignty and the harmonization of the digital market. For those tracking artificial intelligence dynamics, the case is far from peripheral: copyright rules directly affect the ability to use protected content for training LLMs, an area where legal certainty remains elusive.

If the Court confirms that a single country can impose additional obligations on platforms, the domino effect could be disruptive. Each member state could legislate independently, creating a patchwork of requirements that would make it increasingly complex for tech companies to operate with centralized cloud infrastructure. It is not difficult to imagine a scenario where, to manage training datasets with lawfully collected content, enterprises choose to shift processing to on-premise environments, where they can demonstrate precise control over data and its provenance.

Belgium, for its part, defends the law as necessary to ensure fair compensation for creators, an objective made even more urgent by generative AI. With models capable of producing text, images and music from existing works, the question of who should pay for the use of that material becomes central. The legal battle therefore anticipates a broader conflict: on one side, Big Tech’s push to maintain a uniform and predictable regulatory framework; on the other, the demand for stronger protection for rights holders, often backed by national governments.

The EU Court’s verdict, expected in the coming months, will influence not only streaming platforms and social networks, but also those developing and deploying artificial intelligence models. Those pursuing on-premise deployments, seeking data sovereignty and legal risk reduction, could find in this regulatory fragmentation an additional argument in favor of local architecture, where compliance can be managed directly. Meanwhile, the tension between EU law and national autonomy continues to grow, with AI acting as a catalyst.