The AI copyright lawsuits: What happens when machines learn from culture?

29 September 2026facebook iconThe entertainment industry is embroiled in legal battles as the result of the vast development of AI. Record companies, film studios, publishers, authors, photographers and visual artists are suing AI music platforms, image and video-generation companies and other entities training AI systems.

Disney, Universal and Warner Bros. have sued MiniMax and its Hailuo AI platform over alleged unauthorized use of films and characters. Similarly, Getty Images is suing Stability AI, and a class action by visual artists against Stability AI, Midjourney and DeviantArt alleges that billions of protected images were copied without consent or compensation.

Vast sums are at stake. But the legal disputes raise major questions that the current copyright framework is unequipped to address, such as what “copying” a work used to train machines entails, whether a trained model contains the work it learns from and when does copyright infringement actually occur. Most broadly, the fundamental question being raised is who should control the huge cultural and artistic capital accumulated through centuries and now used by AI.

Suno, one of the leading AI music-generation systems, enables users to create a song or music piece from a text prompt. The system can generate a full recording that contains vocals, lyrics, melody, harmony, rhythm and instrumentation.

The model is trained on millions of songs. During training, those songs are converted into mathematical representations and the system sets billions of internal parameters to learn statistical relationships within the data. The result is a statistical system capable of generating new songs from simple text description.

Legally, this distinction is being used by AI companies as part of their legal defense. Their claim is that they train from the existing material rather than creating copies for distribution. The other camp claims that training itself requires copying and that models can retain enough information about particular works to reproduce protected material. For them, AI companies are using their intellectual property to build commercial products that compete with the original material being “copied.”

Now the issue is flooding the courts. On July 31, the Munich Regional Court ruled that Suno had infringed copyright by using songs represented by the German collecting society GEMA to train its system without authorization. The court found that protected works could remain memorized within the model and could be reproduced in response to prompts. Suno has responded by disputing the allegations and further proceedings will follow.

The ruling brings to the fore the claim that AI training doesn’t simply produce abstract mathematical data. Moreover, training is only one of several legal questions.

In music, there are various forms of copyrights. For example, writing a melody and lyrics is one form, while the actual audio recording is another. For a company like Suno, therefore, copyright infringement can occur whether either an existing melody or audio recording is being replicated in a recognizable fashion.

This extends to the reproduction of a vocal performance that imitates closely that of a real singer and involves the possible simulation of a person’s name, voice, likeness and identity. The same goes for musicians, actors and other performers who recognize their work in AI-generated music.

There are therefore many aspects of creative work that are at stake here. But it is important to highlight that these lawsuits do not represent a straightforward struggle between artists and technology companies.

Major conglomerates like Universal, Sony, Disney, Warner Bros. and other studios control enormous recording catalogs, vast libraries of films, television programs and characters. Their goal is to establish that training AI on their catalogs must result in a licensing revenue stream derived from intellectual property exploitation.

Legally, this is a valid position based on private ownership. But the interests of these corporations does not correspond to those of the creative workers whose labor produced the underlying art and culture.

Musicians, actors, writers and animators have virtually no control over their work, the results of which are owned totally or partially by corporations through various legal mechanisms such as work-for-hire contracts. Now AI has made this contradiction impossible to ignore.

Entertainment unions have accommodated themselves to the needs of conglomerates, while workers have strongly expressed their opposition to the use of AI as a means of replacing their labor. During the Writers Guild and SAG-AFTRA 2023 strikes, writers and actors expressed widespread concern and anger over digital replicas, automated production and the destruction of employment. This applies to musicians, voice performers, animators and other workers face the same pressures.

But the contracts the unions negotiated included consent provisions that did little to protect workers. More fundamentally, they left corporate ownership and control of the technology untouched.

This is the contradiction entertainment workers are facing: companies retain full control and unions, accepting this framework, tell workers to accept or even strengthen the stranglehold of the corporations.

Yet, all creative work that trains AI models was created by generations of composers, performers, engineers, animators, writers and more. In music, millions of jazz, blues, classical music, rock, hip-hop or electronic songs produced through decades constitute the knowledge base. In visual art, literature, photography and cinema the same applies. Through individual creativity, artists have first learned from, then enriched a huge social wealth of art and culture available to humanity.

What happened during the 2023 Writers Guild and SAG-AFTRA strikes that shows how the union leaderships handled the fight over AI, and why did the concessions they won fail to protect workers?Can you explain what is meant by socialized production versus private ownership, and how that concept applies to something like an AI model trained on centuries of collective artistic labor?Ask more questions at SocialismAI.comAI makes this mechanism blatantly concrete. No individual human being has the capacity of an AI model to absorb millions of songs or films.

Under capitalism, however, this wealth of knowledge is not simply available free of charge: private property places a tag on it. What is taking place now in courts around the world is primarily infighting within sections of capital that seek to claim ownership or revenues from social knowledge. In other words, it is a struggle within private property.

From the creative worker’s perspective, there are many reasons to oppose the use of AI: under capitalism, it destroys jobs, reduces wages and intensifies exploitation. However, the strategy to defend jobs and living standards demands a head-on struggle against corporate ownership of the cultural wealth created by this labor through centuries, not against a technology that can improve living standards.

This struggle among corporations also raises the question of how artists’ livelihood must be addressed. All creators, whether performers, composers, actors, writers, filmmakers, animators or other cultural workers, have been told, since they began their pursuit of arts, that their profession cannot guarantee a decent living.

Yet, conglomerates stand to extract tens of billions from the redistribution of intellectual property revenues. Artists and cultural workers, like all workers who contribute to the production of social wealth, must be guaranteed secure incomes, healthcare, housing, education, studios, rehearsal facilities and access to the means of production.

AI is no exception. Under capitalism, AI’s ability to reduce the labor required for production becomes a weapon against workers. The owners of the technology appropriate the profit, while workers confront unemployment, cuts and austerity.

Under workers’ control, AI could reduce socially necessary labor, shorten working hours, expand access to education and culture and, importantly, give people greater freedom for creative activity.

This is why workers must see the current lawsuits for what they actually are: a redistribution of social wealth at the top of society, among competing corporate entities and oligarchs. Copyright law is being used to regulate the claims of competing corporations, but it cannot reconcile the contradiction within cultural production. AI has brought the contradiction into the open, raising the question of public ownership of cultural and artistic means of production. Who should controls society’s accumulated artistic, cultural and technological wealth is the fundamental question on the agenda.

Workers and artists share a common interest in opposing the use of AI to destroy their livelihoods. But that struggle must now take on a political form as a challenge to the capitalist relations of production, rooted in private ownership and control of the means of production.