LAION
A German non-profit that publishes very large open training datasets, including the image collection behind Stable Diffusion.
LAION, short for Large-scale Artificial Intelligence Open Network, is a non-profit research organisation based in Germany. It does not build models. It builds and publishes the datasets that models are trained on, which is an unglamorous job that almost nobody else does in the open. Its image dataset LAION-5B was the training material behind Stable Diffusion, and it has since released collections for audio and video too.
That role puts LAION at the centre of the training data argument. Its datasets are assembled by scraping the public web, which means they contain copyrighted work that nobody asked permission for. LAION releases them for non-commercial research and points to a 2024 ruling by the Hamburg Regional Court that allowed it to do so under a German research exemption. Supporters see the only realistic way to keep AI research open and inspectable; critics see industrial scale copying with a research label on it.