Introduction

In last year's contribution the legacy film and television industries were described as 'disrupted and plagued by uncertainty,' and the arrival of AI in the creative industries as the lobbing of a 'hand grenade' into an already combustible mix. One year later, the grenade has arguably detonated. What was, in 2025, a landscape defined by threat, speculation and legal positioning has become, in 2026, one defined by courtroom verdicts, regulatory obligations, platform consolidation, and the sobering realisation that much of what the industry predicted would happen – ie, the decentralisation of content creation and distribution, the commoditisation of production technology, and the erosion of traditional revenue models – has, in large measure, arrived.

Yet the creative industries have not simply capitulated. The past twelve months have, if anything, demonstrated the remarkable adaptability and resilience of storytellers, creative talent, producers, distributors, and their representatives, as they navigate a landscape that would have been almost unrecognisable even five years ago. The questions surrounding what 'content' means – who creates, owns, and profits from it, and how it is created and disseminated, are no longer theoretical. These are the immediate defining issues for the creative industries and policymakers.

Artificial Intelligence – The Battlefield Crystallises

A year ago, it was noted that legal precedent on the 'seminal issues' of AI training and copyright infringement was 'forthcoming shortly.' The past twelve months have delivered the first substantive judicial pronouncements in multiple jurisdictions – unfortunately, not necessarily providing clarity, but at a minimum forcing the debate out of the realm of academic conjecture and into courtrooms and boardrooms where decisions are being made.

In the UK, the long-awaited judgment of the High Court in Getty Images v Stability AI ([2025] EWHC 2863 (Ch)) was handed down by Mrs Justice Joanna Smith on 4 November 2025 – the first UK ruling to examine how copyright and trade-mark law applies to the deployment of generative AI models. As Fieldfisher represented Getty Images in these landmark proceedings, and the claim for secondary copyright infringement is on appeal, commentary on the matter will be circumspect . With respect to copyright, the Court addressed the issue of secondary infringement, and in particular whether Stability's downloading of its model into the UK amounted to importing an 'infringing copy'. While the Court found that Getty Images' copyright-protected works were used in the development of Stable Diffusion (wherever that training did take place), and that an AI model constitutes an 'article' for the purposes of UK copyright law even if it is in intangible form (eg, downloaded over the cloud), it concluded that the model itself did not, on the facts, amount to an 'infringing copy' because it did not necessarily retain a copy of the works in question; however, noting that reasonable minds could differ on whether retention of a copy is a predicate to finding secondary infringement, the Court granted Getty Images permission to appeal on this issue. Getty Images did prevail on its trade-mark claims: where Stable Diffusion generated synthetic images bearing watermarks identical or similar to Getty Images' trade marks, Stability – having control over the training datasets – bore responsibility for any trade-mark infringement, rather than the users who had entered the prompts. The Court expressly rejected Stability's attempt to deflect liability to its users, confirming that control over training data carries with it accountability for the outputs that training data produces.

The Court acknowledged the 'very real societal importance' of balancing the interests of the creative sector and AI development; however, as the decision in the case was, to a degree, fact-specific, there remains significant room for further discussion and argument with respect to fundamental legal questions, as well as potential legislation.

On the legislative front, the UK government has adopted what can only be characterised as studied caution. On 18 March 2026, the government published its much-anticipated Report on Copyright and Artificial Intelligence, required under the Data (Use and Access) Act 2025, alongside an economic impact assessment. Having previously indicated a preference for a text and data-mining exception with an opt-out mechanism for rights' holders, the government formally abandoned that position, declaring – with commendable if somewhat surprising candour – that it 'no longer has a preferred option.' The House of Lords Communications and Digital Committee had already, in its March 2026 report, warned pointedly of the risks that generative AI poses to the UK's creative industries. During an evidence session in January 2026, the Secretaries of State for Science and for Culture conceded that the government's earlier expression of a preferred approach had been, in effect, a mistake – an admission of some consequence in a policy area of this magnitude. In the meantime, over 40 independent publishers co-ordinated letters of claim to leading AI developers in February 2026, alleging the unauthorised use of copyright-protected books and literary works in AI training. The UK continues to consult, engage, and wait, while AI developers, rights' holders, and their counsel operate in what one could describe as a regulatory vacuum that shows little sign of being filled in the near term.

The contrast with the EU's approach could hardly be more pronounced. The EU AI Act's obligations for providers of general-purpose AI (GPAI) models – covering transparency, copyright policy, documentation, and systemic risk management – entered into force on 2 August 2025. The final GPAI Code of Practice, published by the European AI Office on 10 July 2025, provides practical compliance guidance, including an obligation on providers to establish and maintain comprehensive copyright policies applicable to all GPAI models distributed within the EU. By the time this piece reaches readers, the Commission's full enforcement powers – including the power to impose substantial fines – will have entered into application. The extraterritorial reach of the Act means that no AI developer distributing GPAI models in the EU can treat compliance as a matter for another jurisdiction. The EU has chosen to legislate decisively, and the gap between the EU's regulatory posture and the UK's is now substantial and growing.

Naturally, the US landscape also continues to generate judicial activity. In the New York Times litigation against OpenAI and Microsoft, a January 2026 court order required OpenAI to produce a 20-million-entry log of anonymised ChatGPT conversations – a discovery development that suggests the courts are taking the plaintiffs' output-similarity claims seriously. The Disney, Universal and other studios' case against Midjourney proceeds in the Central District of California, a proceeding that may, at last, provide US-based judicial guidance on whether training on copyrighted works, and the resulting outputs, constitutes infringement at scale. Somewhat earlier, a February 2025 partial summary judgment in the Thomson Reuters v Ross Intelligence matter found that the copying of Westlaw headnotes for AI training purposes was not fair use – an early judicial signal, however narrow and fact-specific, that the fair-use doctrine has its limits even in the AI context.

Not all of the significant developments of the past year have taken place in courtrooms. In December 2025, Disney announced a three-year licensing agreement with OpenAI permitting OpenAI's Sora video-generation platform to produce content using over 200 Disney, Marvel, Pixar, and Star Wars characters, in exchange for a reported USD1 billion equity investment and broader use of OpenAI tools across Disney's operations. Whatever one's view of the commercial terms, the deal was a significant proof of concept: two of the most consequential actors in their respective sectors demonstrating that commercial licensing solutions to the training-and-output problem do not require judicial or legislative resolution as a precondition. The rather inconvenient subsequent development – OpenAI's decision to discontinue Sora in March 2026, before the licensed content was substantially exploited, effectively rendering the specific terms of the arrangement moot – serves simultaneously as a cautionary tale about the risks of anchoring commercial strategy to a particular AI platform, and as a reminder of the extraordinary pace at which the underlying technology is evolving. In the music industry, Warner Music Group settled its claims against AI music generator Suno in November 2025, with a new licensed music subscription service anticipated in 2026.

The volume and breadth of licensing activity across the creative industries – together with the proliferation of AI platforms trained exclusively on proprietary or licensed content – confirms that the expectation of remuneration for the use of creative works in AI training is becoming normalised commercial practice, regardless of whether the law presently compels it. AI training clauses in talent agreements, noted in last year's piece as a 'common occurrence,' are now standard drafting. Advisers are increasingly being asked not merely how to protect existing works and existing personas (ie, name and likeness rights), but how to structure new arrangements that anticipate and accommodate the use of those works in AI training from the moment of their creation. Regarding the former, the recently filed trade marks by Matthew McConaughey in the US in audio-visual and audio clips are an interesting example of ingenuity, potentially adding another arrow to the quiver of potentially available claims to protect against misappropriation of one's likeness/voice.

What has perhaps surprised many observers is the extraordinary leap in AI-generated video quality over the past twelve months. Tools capable of producing professional-quality audiovisual content from text and/or voice prompts – building on and in some cases superseding the capabilities of Sora and including platforms such as Google's Veo 3 and SeeDance – have arrived at a pace that has left the production community grappling with questions not merely of copyright, but of employment, creative identity, and the very nature of authorship.

Naturally, the negotiations between the US creative talent guilds and the Alliance of Motion Picture and Television Producers (AMPTP) reflect ongoing grappling with these issues, and, while the Writers Guild of America (WGA) focused its latest effort on having some influence over the licensing of its members' works for AI training purposes, the recently updated Screen Actors Guild (SAG) Basic Agreement reflects the concern of SAG's constituency that the deployment of "Synthetics" will become more commonplace, and SAG's attempt to rein in the use thereof.

TikTok, the Creator Economy, Micro/Vertical-Drama: A Year of Reckoning

Last year's piece noted the growing competition for 'eyeballs' from social media, YouTube, and the 'creator economy.' The past twelve months brought this competition to something of a dramatic head – at least in the United States – with the extraordinary episode of TikTok's near-death and partial resurrection.

To summarise briefly a remarkable regulatory and commercial saga, having been given nine months by Congress to find a non-Chinese buyer or face a ban, ByteDance had not divested its interest by the 19 January 2025 deadline. After the Supreme Court (in a unanimous decision) upheld the constitutionality of the underlying legislation, TikTok went dark briefly – eliciting a response on social media that was, perhaps unsurprisingly, both immediate and emphatic. President Trump, who had previously sought to ban the platform during his first administration, then extended non-enforcement via executive order on multiple occasions throughout 2025, citing ongoing divestiture negotiations. Ultimately, ByteDance signed binding agreements with a consortium of American and global investors for the sale of TikTok's US operations, with ByteDance retaining less than 20% of the new entity, and a compromise resolution was reached with respect to TikTok's proprietary algorithm – the source of its extraordinary commercial power and the root of the original national security concerns.

This saga illustrates that the short-form video ecosystem, which TikTok effectively created at scale, and which has since been replicated by YouTube Shorts, Instagram Reels, and a range of other platforms, is not merely a marketing channel for the traditional entertainment industry. Increasingly, such platforms (separately and certainly in the aggregate), constitute an alternative entertainment industry in its own right, with its own talent ecosystem, creative culture, and revenue models. During the year of uncertainty surrounding TikTok's US future, creators were compelled to diversify their platform presence, brands reconsidered TikTok-first strategies, and alternative platforms benefited materially.

Meanwhile, industry analysts estimate that the global micro (or vertical) drama industry generated roughly USD11 billion in 2025 and could reach USD14 billion in 2026. While the vast majority of such revenue is generated from consumption in China, in the second largest market, the US-based streaming platforms are also experimenting with the same, with varying degrees of commitment. Needless to say, while this new revenue stream is growing quickly, the revenue model for this particular form of content is platform-centric, and challenging for other participants to monetise meaningfully, particularly given the AI-led low cost of production. Moreover, whether the nature of this content can be successfully adapted to broader US/European tastes and preferences is still a matter of experimentation.

However, the creator economy as a whole, estimated in certain industry reports at USD250 billion and growing, has firmly established itself, not merely as a complement to the traditional entertainment industry, but as a genuine competitor for audience attention and creative talent. The most successful creators – commanding global audiences of tens of millions (or sometimes virally reaching hundreds of millions), with diversified revenue streams spanning advertising, merchandise, licensing, live events, and, increasingly, their own production ventures – are negotiating arrangements that would have largely been the domain of only "A-list" creative talent a decade ago. Studio executives are now scrambling to identify the next "Backrooms" and "Obsession", as well as determine whether to revise development departments and strategies. Whether the contractual frameworks developed for the traditional media ecosystem are adequate for these new relationships is a question being asked with increasing frequency, and one that entertainment lawyers will need to answer.

Consolidation: (Silicon Valley Won Before the War Started)

When this publication last surveyed the streaming landscape, the 'streaming wars' were winding down, with profitability having displaced subscriber growth as the primary investor metric. That transition has, over the past twelve months, accelerated into something that looks less like a competitive market and more like a consolidation towards a handful of dominant, vertically integrated entertainment empires.

The SkyDance/Paramount, Netflix, Warner/Discovery saga has been the subject of innumerable articles and analyses, and the story is far from over. For those who remember Netflix's origins as a DVD-by-mail service, and the more recent origins of Skydance, the trajectory of the "disruptors" is vertiginous. Tech companies, tech visionaries, now arguably own (or are likely to own) three of the legacy major studios and comprise two other dominant media companies. These are largely consumer-led companies, using consumer-consumption algorithms and big data to determine what consumers want to see and therefore which content to finance and produce or acquire: in some ways, the opposite of the origins of the major studios, which were led by the stories the studio heads and executives wanted to tell or believed were worthwhile artistically and/or commercially to bring to the public's attention, and which would "find" their audience. The significance and consequences of this relatively tectonic shift in the content creation eco-system are still to be felt.

Interestingly, however, like the trend of consumers who are now purchasing vinyl and turntables, certain aspects of the disrupted eco-system have arguably come full circle. For example, the return of the bundle – in streaming rather than cable form – was, in retrospect, entirely predictable, and is rather ironic. The industry spent the better part of a decade dismantling the cable bundle and celebrating the freedom of the a la carte subscription model and is now engaged in reassembling something that bears a striking family resemblance to it. The Netflix-Apple-Peacock bundle is the most prominent of a proliferating range of digital bundling arrangements designed to address 'subscription fatigue' – itself the entirely foreseeable consequence of the unbundling strategy. Meanwhile, Netflix reported that its advertising-supported tier has grown to 190 million monthly active viewers, supported by a proprietary ad-tech platform launched in late 2025. As advertising-supported viewing becomes the dominant growth engine, the streaming model increasingly resembles traditional broadcast television – which raises the interesting question of whether the creative freedoms that subscription streaming originally enabled will endure as the economics of advertising re-establish their gravitational pull on commissioning decisions.

Similarly, brands' desire to reach consumers via different forms of content, not just traditional advertising and commercials, has brought us back to longer form, brand-sponsored content, and for readers who are too young to know, the reason the serialised radio shows that eventually became a staple of daytime broadcast television in the US were called "soap operas" was because they were overtly sponsored by detergent companies. Back to the future of longer form content "brought to you by" and "presented by"…

Live Events and Experiences – the Counterweight Matures

Last year's piece identified live and immersive experiences as a 'counter-trend' to the increasing digitalisation of entertainment. The past twelve months have confirmed that this is no longer a counter-trend, but a structural component of the entertainment industry's value chain.

Netflix Houses opened, as anticipated, in November 2025: two 100,000-square-foot permanent venues at the King of Prussia Mall in Philadelphia (with another scheduled to open in Las Vegas) and the Galleria in Dallas, firmly establishing the 'phygital' entertainment model – branded, Intellectual Property (IP)-driven, permanent, and capable of being refreshed with new content as the catalogue evolves. Fieldfisher's client, Tokyopop's inaugural Naruto live event in Berlin, which also opened in late 2025, provided a further illustration of the same dynamic: the conversion of beloved anime IP – a franchise commanding a global fanbase of extraordinary depth and loyalty – into a physical, communal experience.

Netflix's pivot to live events is no less significant a development. The streaming of WWE Raw, NFL games on Christmas Day, and a growing portfolio of sporting and entertainment events has fundamentally redefined what a streaming service is – placing Netflix in direct competition with the broadcast and pay TV platforms that live sports has historically sustained. According to one survey, approximately 69% of sports fans in the US now watch at least some live games via streaming platforms, essentially on par with broadcast and cable.

The Las Vegas Sphere continues to serve as the paradigm for a new category of venue: neither concert hall, cinema, nor theme park, but something genuinely novel – an immersive, fully programmable environment capable of enveloping its audience in a controlled sensory world. Sphere Entertainment is actively exploring smaller iterations, suggesting the model may be replicable at a fraction of the original capital cost and therefore deployable in markets where the full-scale Las Vegas installation would not be commercially viable. Universal's Epic Universe, which opened in May 2025, set new benchmarks for theme-park immersion and established what the industry now regards as the baseline for technologically sophisticated experiential entertainment. The International Association of Amusement Parks and Attractions (IAAPA) Expo – the sector's principal trade event – recorded attendance of 38,520 in 2025 and has announced a 50% expansion for 2026, with over 200 companies on the exhibitor waiting list. The message is unambiguous: the immersive experience industry has moved decisively from concept validation to core business.

For the entertainment lawyer, the proliferation of immersive experiences generates advisory work of considerable complexity. IP licensing and the production, management and support of physical experiences involves a fundamentally different matrix of rights from digital distribution and is obviously more akin to the agreements and issues encountered in connection with theme parks. The intersection of AI-generated content with physical environments – adaptation of imagery and works to audience input – all raise fresh questions about authorship, rights clearance, and liability.

Conclusion

Last year's conclusion reflected on storytelling as the primal, enduring element of human creativity – something that technology changes the means of delivering but cannot itself replace. The ultimate measure of success in this industry is whether the story being told is worth telling. From Aristotle's amphitheatre to the Las Vegas Sphere to whatever extraordinary medium or technology or venue comes next, that question has always, in the end, been the only one that truly matters. That said, the context in which this question is answered has become considerably more complex.

The entertainment industry of mid-2026 is one in which the legal framework for AI and copyright in the UK remains unresolved and in active political flux; the EU has enacted compliance obligations that its enforcement apparatus has only just begun to activate; the US courts are advancing, purposefully if slowly, towards verdicts that will set the parameters of the AI licensing market for years to come; ownership of the copyright in an AI-generated work from a US copyright-law perspective requires an analysis of a myriad of factors designed to reach a conclusion as to whether there has been sufficient "human contribution" to the work; and the industry has, in the meantime, proceeded largely on the basis of commercial negotiation rather than legal certainty.

The streaming landscape has consolidated with remarkable speed, and the imperative to secure 'Big IP' that was noted last year has only intensified, as the surviving platforms seek franchise potential capable of anchoring a subscription base, driving a live experience, and justifying a global licensing programme. The creator economy has demonstrated, through the extraordinary TikTok episode, both its fragility – a single regulatory action threatened to erase an entire commercial ecosystem – and its extraordinary resilience. Meanwhile, the live experience sector has matured, decisively, from counter-trend to strategic pillar.

As for what comes next: quantum computing, the continued advancement of brain-computer interface technology (Neuralink's human trials proceed apace), and the gradual maturation of spatial computing all await their moment to reshape the entertainment landscape in ways that are, for the moment, largely the province of science fiction. But then – as remains stubbornly true – the science fiction of yesterday has a reliable habit of becoming the commercial reality of today, and the legal challenge of tomorrow. In this, as in so many aspects of the current media and entertainment landscape, the law is, as ever, running to catch up.