US Court Dismisses Copyright Infringement Case Filed Against AI Image Service Providers

In a major blow to artists in their pushback against AI image generators trained on their art, they have lost their first copyright battle at a US district court.

The class-action lawsuit was dismissed by Judge William H. Orrick, who ruled that it wasn't plausible to determine if AI-generated images were indeed in violation of copyright laws.

The lawsuit in question was filed against three AI image generation service providers - Midjourney, Stability AI, and DeviantArt.

However, the defendants made a motion to dismiss the case, which was largely" granted by the judge on the grounds of the lawsuits being defective.

The plaintiffs, though, have been given the chance to amend their submission on how exactly the said service providers infringed any copyright laws.

Lawsuit Defective on Numerous" Counts, Claims Judge Orrick

The lawsuit was filed by artists Karla Ortiz, Sarah Anderson, and Kelly McKernan, who claimed that the LIAON datasets used by the AI service providers included their work. Thus, training their AI models using these datasets was in violation of copyright laws, the artists contested.

However, Judge Orrick pointed out that their argument had numerous" defects - including the fact that McKernan and Ortiz hadn't filed any copyrights for their artwork with the US Copyright Office.

As for Anderson, only 16 of the hundreds of images she cited in the lawsuit had actually been copyrighted. She will be allowed to pursue her copyright infringement claim for these 16 artworks.

The other problem for plaintiffs is that it is simply not plausible that every Training Image used to train Stable Diffusion was copyrighted (as opposed to copyrightable) or that all DeviantArt users' Output Images rely upon (theoretically) copyrighted Training Images, and therefore all Output images are derivative images.Judge Orrick

Orrick did offer plaintiffs an opportunity to amend their lawsuit. However, he refused to believe that copyright claims based on a derivative theory can survive absent substantial similarity' type allegations" even if the artists provided more clarity and narrowed down their allegations to output images drawn upon training images based on copyrighted art.

According to the judge, cases that the artists have been relying on recognize that for a copyright infringement claim to be valid, the derivative work by the alleged violator must contain protected pieces of art from the original work or bear similarities to it.

Implications of the Ruling

The court ruling potentially implicates that unless artists can prove that an AI-generated image has been referenced directly from their work, such copyright cases might not stand.

The fact that artists often do not get many of their artworks copyrighted might leave AI service providers off the hook, too.

This is admittedly a massive blow to artists who have been campaigning against the practice of training generative AI models on their creative work.

The ruling might act as a precedent for other similar ongoing trials in the US and potentially tip the scale in favor of the tech companies in some cases.

For instance, coders filed a lawsuit against Microsoft-owned GitHub, claiming that codes posted using open-source licenses were to train GitHub's Copilot coding assistant.

The case, filed at a San Francisco district court, accuses Copilot of ignoring, violating, and removing licenses.

However, it's not all bad news for those waging copyright battles against generative AI. Following the example of the ruling, the class-action lawsuit against OpenAI and Sarah Silverman's case against Meta might potentially see the rulings being in favor of the plaintiffs.

In both cases, the respective plaintiffs already have copyright protection.

Overall, it appears that the motion to dismiss copyright claims or suits by defending technology companies can be denied if the alleged original work is copyrighted.

The post US Court Dismisses Copyright Infringement Case Filed Against AI Image Service Providers appeared first on The Tech Report.