GenAI in media: Bots, back catalogues, and battles

Many people still find great charm in US animator Ub Iwerks’s drawings for the 1928 black-and-white Disney film Steamboat Willie, which introduced Mickey Mouse to a world that has remained grateful in the century since. But the film is, in terms of sophistication, worlds apart from even the earliest Pixar feature-film efforts such as Toy Story (1995).

The development of animation is just one example of the constant state of flux, driven by technology, across the media and entertainment industries. In other words, while the actual technology is always new, the fact of technological change is extremely old. And many of the advancements we have seen are generally accepted as hugely beneficial.

The most important tech impact at the moment, in media and entertainment, is exerted by generative artificial intelligence (GenAI). What effect will it have?

The mouse meets machine learning

The scale of what can be achieved, together with the speed and cost efficiencies it offers, amounts to an unparalleled step change. Its potential isn’t confined to elements of production such as special effects. It can also be applied to the process of coming up with ideas, by testing multiple scenarios quickly and cheaply and combining content from different sources. Think Mickey Mouse meets Remy, the genius rodent from Ratatouille, for example.

In 2025, The Walt Disney Company announced that it had licensed certain content to OpenAI to enable the creation of new short fan-inspired video content using OpenAI’s Sora video-creation engine. That agreement has since been discontinued following OpenAI’s decision to shift its focus away from Sora. But the principle is now established, and we’d expect Disney to pursue alternative licensing deals of a similar nature.

From an investment perspective, this may favour companies with valuable intellectual property and extensive content libraries, particularly where management can use AI to open up new licensing and revenue opportunities.

Back catalogues go bespoke

This idea of mashing up content may resonate even more strongly with music, not only when driven by rights holders but also when it is placed in the hands of consumers. Imagine being able to create multiple versions of the same song in different languages or genres, or to combine artists in unexpected ways. Say, a Spanish-language rap version of the Queen classic Bohemian Rhapsody sung by the velvet-voiced Johnny Cash and disco legend Donna Summer.

This functionality reflects the music industry’s efforts to better monetise fandom through premium subscription offerings, in contrast to the historical one-size-fits-all model. At its recent investor day, Sweden’s Spotify, the market-leading music streaming service, announced a licensing agreement with Universal Music, the largest music rights holder, to offer precisely this type of feature.

The radical reimagining of content, personalisation, and localisation offer powerful possibilities. Of these, personalisation is particularly useful in advertising, by making it more effective on more people. The old advertising problem – “half of the money I spend on advertising is wasted, and the trouble is I don’t know which half” – has become less severe in the digital age and could be diminished further still.

This could create opportunities for businesses able to deepen user engagement and charge premium prices, especially where personalisation helps turn passive audiences into higher-value subscribers.

The Dumbo-sized elephant in the copyright room

We see ample opportunity for media companies to exploit GenAI to their advantage. Yet investor sentiment is currently more negative than this might suggest. Why? The reason is – to evoke one of Mickey’s Disney stablemates – the great, Dumbo-sized elephant in the room: the worry that the value of their intellectual property will be diluted, and with it their earning power.

This is where the investment case becomes more complicated, because the same technology that can create efficiencies may also undermine the economic value of the assets that media companies rely on.

In the world of GenAI, anybody can 'create' content – images, videos, music and more – with a few clicks of a mouse and at little cost. The difficulty is that content is produced using GenAI models that have been trained on vast quantities of existing materials, much of it expensive to create, subject to copyright and used without licence – that is, without permission or the payment of any royalty.

Flooding the media ecosystem with such content tilts the playing field against the owners of the original content and competes with it for consumer attention and, ultimately, revenue. On the face of it, that may seem plainly wrong. But whether one can legally do this, without licence, is still a grey area.

Old tunes, new tricks

Music has always borrowed from what came before. Sometimes the influences are overt, sometimes subtler. Judging whether something is original is highly subjective – leading to numerous legal disputes over alleged plagiarism.

The media and entertainment industries are currently in the throes of a great deal of speculation over the impact of GenAI on them. Particularly surrounding copyright, where the legal position was already murky in the pre-GenAI era, and is now even more treacherous.

Famously, a court ruled that ex-Beatles George Harrison’s My Sweet Lord was melodically almost identical to American girl groups The Chiffons’ He’s So Fine, and that Harrison had copied it, albeit subconsciously. By contrast, defendants from Led Zeppelin to Katy Perry have won where courts judged the similarities insufficient.

US copyright law contains a ‘fair use’ provision. In some circumstances, this allows copyrighted material to be used without permission if the new work is deemed ‘transformative’. In other words, the artist has been inspired by one piece of work and used it to create something very different. This defence can be hard to apply, partly because the test is so subjective, and also because courts look at a ‘market impact’ test – whether the new use could reduce demand for, or the licensing value of, the original work or harm competition.

There have already been a number of disagreements between major record labels and AI companies over the use of their copyrighted material to train models and create ‘new’ content. Some of these have been resolved out of court with licensing agreements; others rumble on. Universal Music estimates that AI-generated music accounts for less than 1% of listening on streaming services, and its contracts increasingly exclude unlicensed AI music from the pool of streaming royalties paid to rights holders.

Nevertheless, investors would welcome greater legal clarity: GenAI could create new licensing, personalisation and efficiency opportunities, but also erode the scarcity, pricing power and intellectual-property protections on which much of the media and entertainment companies’ value still rests.