Süddeutsche Zeitung1 hour agoTech1 min readGermany

Federal Court of Justice is negotiating training data for generative AI

Quick Look

The first civil senate of the Federal Court of Justice has negotiated a database with 5.85 billion text-image pairs that will be used to train generative AI models.
The negotiation focused on the legal framework for collecting and processing this data.

AI-generated summary

Why It Matters

The Federal Court of Justice negotiated a database with 5.85 billion text-image pairs that is intended to be used to train generative AI models. The data is discussed in connection with the use of such AI systems.

Font size

Artificial intelligence has now reached the Federal Court of Justice (BGH). Although that's not entirely correct, because this Thursday the first Civil Senate negotiated a database that is initially only intended to provide the material for "training" generative AI: 5.85 billion text-image pairs. But of course the collection and processing of data is clearly reflected in the light of the AI models whose use they are supposed to make possible.

Open Questions

What legal basis applies to collecting training data for AI?
How is copyright protection guaranteed when using text-image pairs?
What consequences could the ruling have for AI development in Germany?

Federal Court of Justice is negotiating training data for generative AI

Quick Look

The first civil senate of the Federal Court of Justice has negotiated a database with 5.85 billion text-image pairs that will be used to train generative AI models.
The negotiation focused on the legal framework for collecting and processing this data.

AI-generated summary