Output classification assigned responsibility along two legal paths
The Munich Suno ruling classifies reproduction and provider responsibility, while Louisiana criminalises creating prohibited synthetic content; the paths concern different acts, harms, and procedures.
Artificial Intelligence··Midday
The Munich court treated six outputs as reproductions
Munich Regional Court I classified the relationship between six compositions and model outputs under copyright law in case 42 O 763/25 between GEMA and Suno. The court found that the disputed works had been memorised in model parameters and could be retrieved as substantially similar outputs through simple prompts containing a song title, lyrics and style. It treated that result as reproduction under section 16 of German copyright law and declined to apply the text-and-data-mining exception. The ruling also said Suno used stream-ripping methods to obtain music from YouTube and circumvented a technical protection measure. For outputs created from open-ended prompts, it placed responsibility for determining the content primarily with the provider rather than the user. The court largely granted GEMA's claims for an injunction, disclosure and damages. This is a civil judgment based on the specified works and evidence; it is not final and remains open to appeal.[1]
Louisiana criminalised creation through a prohibited-content category
Louisiana's Act 892, effective from 1 August, establishes a different legal classification. It criminalises the creation of AI-generated child sexual abuse material. Its scope is not confined to a displayed file: the definition also includes undeveloped film, videotape, and data stored on a computer disk or electronically that can be converted into a visual image. State Representative Delisha Boyd of New Orleans, a co-author of the measure, told Fox 8 that AI was advancing rapidly and that the law was intended to stop the creation of those images. The report describes the measure as one of several state laws taking effect that day. It does not provide penalty amounts, name the authority responsible for enforcement, or describe a case brought after commencement. The responsibility path here begins with the act of creating a prohibited category of synthetic content, rather than a similarity test involving identified protected works.[2]
A shared technology label does not merge the legal paths
AI output is the object of legal assessment in both developments, but the classified act and structure of responsibility differ. In the Suno judgment, the court treated retrieval of six identified works from the model as reproduction; the training method, circumvention of technical protection, openness of prompts, and provider's role in determining content were considered in the same case. Louisiana's law does not examine copyright similarity or model training. It places creation of synthetic material defined as child sexual abuse content into a criminal category and extends that category to stored data convertible into an image. The first is a non-final civil judgment addressing claims for injunction, disclosure, and damages. The second is a criminal prohibition now in force, with no enforcement case or penalty detail reported in the article. This distinction locates responsibility without equating the harms or gravity of the events: one path concerns the work-output relationship and provider role, while the other begins with creating a prohibited content category. The sources also describe the next step only narrowly: the Suno judgment may change on appeal, while the Louisiana report identifies neither an enforcing authority nor a first case. That procedural distinction matters. Each path will proceed through its own legal process.[1], [2]
Related columns
For more information on this topic, you can read the related columns.
