Judge finds artists' copyright claims against AI image generators partially viable but requires clearer pleading; inference-time name-based infringement unlikely to succeed

A US federal judge ruled that artists' direct copyright-infringement claims against Stability AI for incorporating their registered works into training data can proceed past the motion-to-dismiss stage, but found unclear pleading across multiple defendants and rejected claims that AI systems infringe copyrights when generating images based on artist names alone. The judge indicated artists must clarify which defendant trained which model infringing-ly and have access to source code to strengthen factual allegations. This ruling establishes that training-data IP infringement is justiciable and requires clear disclosure, but raises the bar for proving individual work inclusion in datasets of billions of compressed images.