AI and copyright: music

In our last article, we saw that new types of pirated content have been appearing recently: fake books and fake audiobooks. Music has long been the poster child for anti-piracy measures, thanks to a well thought-out business model that seemed to satisfy the vast majority of users. But recent developments in AI could change all that…

Music, a poster child for anti-piracy

Music was the first industry to be hit hard by piracy. Even before the widespread use of personal computers, bootleg tapes and ripped CDs were swapped under the table. They were usually recordings taken from concerts or rehearsal sessions, or sometimes artisanal copies of existing albums. Then, in the early 2000s, MP3 sharing exploded with famous sites like Napster. The music industry was losing revenue, but it adjusted quickly. Solutions that allowed piracy to be reduced enormously were found. Platforms with competitive offers gave users the impression that they could find almost anything they wanted on Spotify or Youtube music without having to multiply subscriptions; for such offers, they were willing to pay. Over the last few years, music piracy has taken a smaller and smaller share of the industry’s revenue, and while it has not decreased, it has stabilized at an acceptable level.

AI and Musical Creation

So what can AI automate, and how is this automation relevant for music piracy ?

AIs can greatly improve the quality of pirated copies, transforming low-quality recordings into higher-quality versions, making pirated copies more attractive to consumers. The infamous “bootlegs” of the 80s are resurfacing, with remastered versions that generate no revenue for either artists or record companies.

More specifically, the following uses are possible:

Hijacking Copyright Protection Systems

AI-based systems can subtly modify music files to trick the audio recognition technologies used by streaming platforms and file-sharing sites.

Automatic Distribution

Bots and automated networks can be used to distribute pirated files on a massive scale, making it more difficult to fight against the illegal distribution of music.

AI Generation and Manipulation

AI algorithms, particularly those related to audio signal processing, have enabled significant advances in music creation. Tools such as Jukedeck, Amper Music, and OpenAIs useNet can generate original musical compositions by imitating various styles and genres.

AI technologies also enable advanced manipulation of existing music tracks. Source separation algorithms, for example, can isolate the different components of a piece of music (vocals, instruments, etc.), facilitating the creation of remixes or alternative versions.

These tools can be used creatively, notably in electronic music, but also to circumvent copyright, by slightly modifying tracks to make them unrecognizable to traditional detection systems, while taking advantage of a well-known artist’s reputation to sell pirated content.

The diminishing value of music

As with literature, there is a danger that these musical manipulations will diminish the perceived value of music. Ease of access and the abundance of creations of varying quality can lead users to question why they should continue to pay for music.As with literature, the variable quality of content produced using these new technologies inevitably harms authors, either by creating unfair competition or by devaluing the content.

The reality is that many bands have adapted to piracy and that they rely far more on concert tours and on-site merchandising than album sales to generate revenue.

Reactions from governments and the music industry

Faced with such challenges, the music industry and legislators are seeking solutions to protect artists’ rights and combat piracy by developing anti-piracy or detection technologies (which are often quickly circumvented). Music lobbies and governments are seeking to introduce tougher legislation, but this too is moving at a slower pace than technology. Consumer education continues, but its effectiveness remains limited.

The RIAA against AI companies

Recently, this evolution in music creation was illustrated by a lawsuit by the Recording Industry Association of America (RIAA), which brings together the big names in the music industry such as Warner and Sony music, against two AI music-generating companies, Suno and Udi, in June 2024. The plaintiffs accused the generators of copying popular (and still under copyright) recordings, incorporating them into AI models and then generating tracks of comparable quality to human recordings.

The two AI companies’ response is almost identical: they have created music that did not exist, and could not have existed in their intervention. This music can’t, therefore, be considered as copy. AIs simply took inspiration from existing things and styles, which are nobody’s property, just as a human creator does, using technologies such as neural networks.

AI models collect statistics on the exact structure of a song, on how drums or guitars will be used in different styles… The two accused companies also defend themselves by claiming that the analyzed songs are not stored in databases.

The plaintiffs, on the other hand, present these creations as “copies”.

Who’s right? For the moment, the complaint has been rejected by the court; but this case highlights new issues, which will certainly be of prime importance in the years to come.

Join us in September for a new topic. In the meantime, search and removal remain the most effective methods for getting content off the web as much as possible. If you have a film, series, software or e-book to protect, don’t hesitate to call on our services by contacting one of our account managers; PDN has been a pioneer in cybersecurity and anti-piracy for over ten years, and we’re bound to have a solution to help you. Enjoy your reading, and see you soon!

Share this article

INVESTIGATION AND
TAKEDOWN

LIVE EVENT
PROTECTION

BRAND PROTECTION

IPTV ANALYSIS