Germany's highest civil court, the Bundesgerichtshof (BGH), is tackling a question that has divided the AI industry for years: Can copyrighted works be used to train AI models without asking the creators first? On Thursday, the court will hear a case between a professional photographer and Laion, an organization that provides training datasets for AI image generators. The answer could have far-reaching consequences for innovation in Germany and Europe.
The essentials
- Plaintiff: A professional photographer is suing Laion, a provider of training datasets for AI image generators
- Scale: The dataset in question contains 5.85 billion text-image pairs and is used worldwide for AI training
- Previous rulings: Lower courts dismissed the case, confirming that such data collection can be lawful under certain conditions
- Central question: Does the "text and data mining" exception apply to commercial AI development without explicit creator consent?
The dispute: Automated data collection vs. copyright
Laion provides massive datasets used globally as the foundation for image-generating AI models. The plaintiff argues that the organization captured, stored, and distributed his photographs without permission—a clear copyright violation in his view. The legal dispute centers on whether automated capture and storage of image data from the internet is covered by existing copyright exceptions.
The industry is intensely debating whether the so-called text and data mining exception—a provision in EU copyright law—also applies to commercial or research-oriented AI development when affected creators have not explicitly consented to their work being used.
What's at stake
The BGH decision is being watched closely by experts and industry representatives. Many AI companies rely on access to large, publicly available datasets to train their algorithms. A court ruling against Laion could have far-reaching consequences for the availability of training data and thus for innovation speed in this sector.
Meanwhile, industry associations and copyright advocates have long called for clear legal boundaries and compensation models. They argue that artists and photographers deserve compensation when their work is used commercially for AI training.
Timing: EU AI Act raises the pressure
The hearing comes at a critical moment. The EU AI Regulation has been directly applicable since August 2024, imposing new requirements on companies to document and ensure the legal soundness of their training data. A BGH ruling against Laion could tighten these requirements further, forcing developers to fundamentally rethink their data sources.
What this means for you
If the BGH holds Laion accountable, German and European AI companies will need to reconsider their training practices—possibly requiring explicit licenses or opt-out systems for creators. This could slow development but also lead to fairer models. For photographers and artists, it could mean stronger protection of their rights. The decision will reveal whether Europe is taking a copyright-first or innovation-first approach.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
