Associate
Generative artificial intelligence (AI) is a branch of AI focused on the creation of original content, simulating human creative capacity. This technology can generate a wide range of outputs, including text, images, musical compositions, sounds, videos, and even lines of code.
It works through sophisticated machine learning models known as generative models. These models do not simply replicate the data they were trained on; instead, they learn the patterns, structures, and styles embedded in vast volumes of content previously created by humans.
During the training process, the AI analyzes how different elements of content relate and are organized, for instance, how a sentence is constructed, how colors blend in an image, or what sequences of notes are typical in a melody. Based on this understanding, the AI can produce original results that follow similar logic, generating content that is novel yet coherent.
One notable case that highlights the growing legal tensions surrounding this technology is the lawsuit filed by The New York Times against OpenAI and Microsoft in December 2023. The newspaper claimed that its articles were used, without authorization, to train language models such as ChatGPT. According to the complaint, this has led to the generation of content closely resembling the newspaper’s original publications, thereby impacting its intellectual property and business model. Although this case pertains to journalism, its implications reach into other creative domains, including music. Today, generative AI is transforming the way music is composed, produced, and consumed, offering exciting possibilities but also raising complex issues regarding the use of protected works. Notable applications of generative AI in music include:
- AI that creates music based on text or descriptions: These tools allow users to input phrases such as “a soft violin piece in the style of Mozart”, and the AI generates a composition accordingly.
- AI that develops user-provided musical ideas: A melody or chord progression can be uploaded, and the AI expands or reinterprets it.
- AI trained on large music catalogs: By analyzing millions of works, the AI learns musical structures, genres, styles, and harmonic combinations, enabling it to create new compositions modeled on that complexity.
While each type of music AI serves different purposes and audiences, they all share a common characteristic: they are trained using third-party works, which, so far, have not been clearly or consistently compensated. The New York Times v. OpenAI case underscores the need to establish clear and transparent licensing mechanisms. Technologies that rely on protected content should pay for its use, respecting copyright laws.
In the context of music, this could mean that companies developing AI music generators may be required to compensate collective management organizations or guild associations, much like other forms of use of copyrighted works.
Although regulation in this area is still in its early stages, it is essential to start reflecting on and addressing these issues. Technological development is moving rapidly, and it is crucial to be prepared so that when more rigorous regulations become necessary, a clear and fair framework is already in place, one that protects creators while promoting responsible innovation.