The judge ruled that DeviantArt and Midjourney had not directly infringed upon Andersen's rights as they did not scrape any images or train the AI software themselves. Instead, the data was collected by non-profit organization LAION, under the direction of Stability, who then developed the AI software. The judge also questioned whether every image in the training dataset was copyrighted and if all AI-generated outputs are unlawful derivatives of images. The plaintiffs' lawyers confirmed they would be filing an amended complaint.
Key takeaways:
- A judge has dismissed copyright infringement claims against DeviantArt and Midjourney, but allowed a case against Stability AI to continue.
- The case was brought by three artists who accused the companies of using copyrighted artwork without permission to build AI tools.
- The judge dismissed claims made by two of the artists as they had not registered their work with the US Copyright Office, but allowed the case brought by Sarah Andersen to continue against Stability AI.
- Judge Orrick advised the artists to clarify their cases against DeviantArt and Midjourney, and expressed doubt over the claim that all AI-generated outputs are unlawful derivatives of images.