Music has evolved over the centuries, from the first bone flutes (dating back to at least 38,000 BC) to the invention of the piano in Italy in 1700. The theremin, invented in 1920 by a Russian physicist, is played by a performer moving their hands near two antennas. However, it was only in more recent times that machines moved from being instruments to actually creating music. The Illiac Suite for string quartet was created by a computer algorithm in 1957. The first album created by an AI was in 2018, “I AM AI”, produced by singer Taryn Southern. However, it is the widespread adoption of generative AI since late 2022 that has led to an explosion of AI-generated music. This technology can generate instrumentals but also vocals, using applications like Udio and Suno AI. Neural networks can be trained on libraries of music and analyse rhythm and melodies to produce original content. This works in much the same way that large language models (LLMs) can produce new text content based on the synthesis of vast amounts of textual training data. Just as LLMs generate text by predicting the most probable next word (or “token”) at a time, musical generative AIs predict the next note or audio token. Music can be generated in the style of existing artists, such as the 2016 song “Daddy’s Car”, notionally in the style of the Beatles, created by Francois Pachet of Sony Computer Science Laboratory in Paris.
The fact that generative AI can create music in the style of existing human artists creates obvious intellectual property issues in an industry that depends on copyright
and digital rights, and paying royalties to artists. Indeed, the first copyright law was the Statute of Anne in 1709. In 1793, musicians were granted the same rights as authors for copyright in France, and French composer Ernest Bourget won a lawsuit conducted between 1847 and 1849 to collect royalties for his music being played at a café. Recording a professional musical track at a studio is an expensive proposition, but now anyone with a computer or phone can type in a few prompts and have a song written. Musical jingles, film soundtracks, and complete albums of music can now be written by AI. So, what does the advent of widespread AI-generated music mean for the industry?
The song “Walk my Walk” by AI artist Breaking Rust was the top of the Billboard Country Digital Song Sales chart in November 2025, with three million streams on the Spotify platform in under a month. AI music has clearly struck a chord with many listeners. Spotify has around 100 million tracks in total, and to give you a sense of the scale of AI-generated music, it claimed in September 2025 to have removed 75 million AI tracks, mostly for impersonating existing musical artists. The Spotify platform allows AI-generated music if it is clearly labelled as such.
The advent of AI music is a seismic event in the music industry. It is estimated that in 2024, AI music generated to the tune of $560 million in revenue. A third of music uploads to music platform Deezer are now AI-generated, and that is just the proportion flagged by that company’s AI detection system. The true proportion may be higher. In a survey that the same company carried out of 9,000 people across eight countries, 97% could not tell the difference between human music and AI-generated music. The International Confederation of Societies of Authors and Composers published a study in December 2024 that music creators will see at least a fifth of their revenues at risk from AI by 2028, as AI-generated music increasingly cannibalises the market and absorbs royalties that would otherwise have gone to human artists, while other AI-generated music deliberately uses their work without compensation.
Since all generative AI is trained on human music, and since that music is protected by copyright, is this fair use or theft? The conflict has been music to the ears of lawyers, and numerous AI copyright cases are rumbling slowly through the legal system. Studios like Universal, Warner and Sony are suing AI companies like Suno, Udio and Anthropic. Anthropic has already settled a high-profile (text) copyright lawsuit (the case Bartz v Anthropic) for $1.5 billion. In November 2025, Warner Music Group and Udio settled a copyright case. Indeed, both parties then appeared to change their tune, announcing after the settlement that they would jointly work on a new AI song creation platform. Many more cases are pending, and AI companies may soon have to face the music in court.
Some human artists have decided to march to the beat of a different drum and to work with, rather than against, AI-generated tunes. Some use AI for assistance in songwriting, or to produce remixes. It is still unclear how AI will be integrated into the existing music business, as artists get to grips with its capabilities and as the legal ramifications of copyright protection proceed through the courts. However, it is certain that music is another industry that is being profoundly affected by AI.
