This post discusses the German case Kneschke vs. LAION, the first key ruling addressing the application of TDM exceptions to AI training. The decision clarifies the practicalities of reproducing protected works for building training datasets while leaving certain issues open.