German photographer sues nonprofit over copyrighted images in AI training dataset
A German professional photographer is suing the Hamburg nonprofit organization Laion at Germany's Federal Court of Justice over the inclusion of copyrighted photographs in a dataset used to train AI image generators. Laion provides a freely available dataset containing 5.85 billion text-image pairs created by downloading and analyzing images from the internet, and lower courts initially rejected the photographer's copyright infringement claim by citing exemptions for text and data mining.
What happened
A German professional photographer filed suit against the Hamburg nonprofit organization Laion at the Bundesgerichtshof (Federal Court of Justice) over copyright infringement related to a dataset used for AI training.
Laion provides a freely available dataset containing 5.85 billion text-image pairs for AI training. To create this dataset, Laion downloaded and analyzed images from the internet to filter image addresses and verify that text descriptions matched the images.
The photographer claims the downloading and analysis of his copyrighted photographs for the dataset constitutes copyright infringement. Lower courts in Hamburg had dismissed his claim, citing exemptions in copyright law for text and data mining that permit automated analysis of digital works when extracting information about patterns, trends or relationships.
Index analysis
The case addresses the question of whether copyright law exceptions for text and data mining apply to the creation of datasets used to train AI image generators.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.