Entertainment Law 2023–2026: What Actually Changed and Why It Matters

Entertainment law has always been about who owns what and who gets paid. What has shifted in the last two years is the more fundamental question underneath that one: who captures value when content gets repackaged at scale by platforms, intermediaries, and now by generative AI?

Three threads have driven most of the consequential developments. Copyright's boundaries are being redrawn by Supreme Court guidance on fair use and licensing markets. Trademark and publicity rights have become the frontline tools for disputes that traditional copyright was not built to handle. And platform liability continues to turn on safe harbors and knowledge standards that are older than the businesses now depending on them.

Here is what practitioners and business people actually need to know.

Copyright: the transformative use argument got harder

For years, the entertainment industry treated "transformative" as something close to a get-out-of-jail-free card in fair use analysis. Add new meaning, commentary, or expression, and the secondary use was probably defensible. That framing still exists, but it is no longer safe to rely on when the secondary use competes in the same licensing market as the original.

The Supreme Court's 2023 decision involving the Andy Warhol Foundation and photographer Lynn Goldsmith made this concrete. The Court evaluated fair use through the lens of commercial purpose and market substitution, and held that where a new use shares substantially the same purpose and occupies the same commercial space as the original, the first fair use factor can weigh against the defendant even if the secondary work carries a different message or artistic intention. The part that should change behavior: "transformative meaning" is no longer a trump card when you are functionally selling the same type of license the rightsholder sells.

For music, film, photography, and digital art, the practical effect is straightforward. Licensing for uses that land in the rightsholder's traditional markets needs to be more explicit. Internal review for uses that might be defended as commentary but are actually deployed as commercial substitutes needs to be sharper. And deal terms need to address downstream licensing more carefully, including who can sublicense into editorial or marketing placements.

This matters even more in an AI context. Many AI-driven outputs are, commercially speaking, designed to substitute for human-created content in the same markets. The Warhol decision does not resolve AI training questions directly, but it strengthens the argument that the relevant analysis is not just whether the output is new expression, but whether the defendant is functionally taking a license the plaintiff would have sold.

A separate 2024 Supreme Court decision involving Warner Chappell Music added another layer of exposure. The Court held that the three-year copyright limitations period does not impose a categorical three-year cap on recoverable damages if the underlying claim is timely filed. For entertainment companies, that changes the math on historical infringement disputes. Catalog owners and publishers will care more about long-range revenue histories. Platforms and distributors will want cleaner evidence of when they received claims and what they did with them. Settlement valuation in copyright disputes may no longer be anchored by a simple three-years-of-damages frame.

Platform liability: safe harbors reward process, not just good intentions

The DMCA safe harbor framework remains the central defense for platforms hosting user-generated content, but two things are worth understanding clearly.

First, safe harbor protection depends on specific knowledge of infringing activity, not generalized awareness that infringement probably happens on your platform. Willful blindness can substitute for actual knowledge in the right circumstances, but the standard is not "you should have known it was out there somewhere." Platforms that invest in repeatable, documented notice-and-action workflows are in a materially better position than those that rely on scale as a defense.

Second, rightsholders have their own obligation on the front end. The Ninth Circuit's decision in Lenz v. Universal Music established that rightsholders must consider fair use before sending DMCA takedown notices, and that misrepresentation claims can follow when they do not. In 2024 and 2025, that matters more because user-generated content is increasingly monetized, automated notices are common, and disputes about short clips, reaction content, and audio snippets are constant. Scale does not eliminate the need for a documented decision process, especially when mass enforcement creates counterclaim risk.

For platforms, the lesson from inducement doctrine is also still relevant. When a service is designed, marketed, or operated in a way that encourages infringement, liability can follow the intent. This remains a live theory for AI-enabled music and video generation platforms where plaintiffs argue that the business model depends on infringing inputs or outputs.

Trademark: parody is not a special pass when you are using the mark to sell something

Entertainment and consumer products now operate in the same social marketplace. Artist merch, brand collaborations, film tie-ins, and parody goods all live together, which means trademark doctrine is increasingly central to entertainment risk.

The Supreme Court's 2023 Jack Daniel's decision clarified something that a lot of creative industry clients had been assuming the wrong way. The Rogers test, which had provided a First Amendment-friendly framework for expressive uses of trademarks, does not apply when the defendant uses the mark as a source identifier for their own goods, even if the use is humorous or parodic. Parody can still be considered, but it gets evaluated inside ordinary trademark likelihood-of-confusion and dilution analysis, not through a special expressive-work gateway.

For merch drops that riff on a brand's trade dress, parody packaging used to sell physical products, and social-first products where the joke is also the brand, this means the creative intent does not change the legal framework. Standard trademark review applies.

A related doctrinal point worth understanding: trademark law has real limits as a credit or attribution mechanism. Attempts to use unfair competition claims to address what is really a complaint about copying creative expression or failure to give credit will run into the principle that trademark law addresses source confusion, not authorship. If the real grievance is about copied expression or missing attribution, the right framework is copyright, not the Lanham Act, unless there is genuine source confusion or a specific endorsement problem.

Publicity rights: the most active frontier

The most reader-relevant shift across the full period is not a single new doctrine. It is acceleration. Publicity rights are increasingly being used to police AI voice cloning, digital doubles, and endorsement-like impressions, and the existing doctrinal tools are better suited to these disputes than many people realize.

Three older Ninth Circuit decisions have become the primary playbook for AI voice and likeness disputes. In Midler v. Ford Motor Co., the court held that deliberate imitation of a distinctive, widely known singing voice for advertising is actionable. In White v. Samsung Electronics, the court recognized that a common law claim can exist where an ad appropriates a celebrity's identity without using their literal name or image, and allowed a false endorsement theory to proceed based on overall impression. In Waits v. Frito-Lay, the court reinforced liability and damages principles for soundalike advertising and recognized false endorsement theories alongside them.

In 2024 and 2025, these cases are being used as direct analogies for AI-generated vocals in the style of a well-known performer, synthetic voices in ads designed to evoke a specific celebrity, and content that implies endorsement through an identity cue rather than a literal depiction. The fact that the technology is new does not change the underlying claim structure. The question is still whether the defendant appropriated the economic value of someone's identity without permission.

For video games and interactive media, where depictions can replicate a person's appearance and context with extraordinary fidelity, two circuit court decisions involving college athletes and Electronic Arts established that realistic depiction of an individual in the same setting that created their commercial value is not sufficiently transformed to defeat a publicity claim. This maps directly onto AI avatars and digital double uses. The more faithful the recreation, and the more it replicates the context in which that person built their value, the higher the legal risk.

California's transformative use test, developed in cases involving celebrity portrait merchandise and comic book depictions, draws the same line. Stylization and genuine creative transformation reduce risk. Photorealistic or literal uses intended to monetize identity are more likely to face challenges. States are also updating their statutes, with New York and Tennessee having enacted provisions that specifically address digital replica concerns, and California labor law now reflecting a policy focus on AI and performer protections that will influence contract drafting even beyond California projects.

What this means for deal terms in 2026

Across all of these developments, courts have become less willing to let broad speech or transformation narratives override the commercial structure of a market. The Warhol decision does this for copyright licensing markets. The Jack Daniel's decision does it for trademark parody goods. The athlete-likeness decisions do it for realistic identity depictions.

At the same time, procedural compliance still provides real protection for platforms. The DMCA framework rewards those that build reliable notice-and-action systems and punishes those that rely on scale to avoid specific knowledge.

The growth area for disputes is identity, not just content copying. Publicity and false endorsement theories built around voice, likeness, and evocation are well positioned for the controversies that AI is generating because they address the appropriation of commercial value directly, without requiring the plaintiff to fit their claim into a copyright framework that was not designed for it.

For creators and talent, define AI boundaries explicitly in every agreement. Specify whether the counterparty may create digital replicas, voice doubles, or synthetic performances. Build in approvals for advertising uses and anything that resembles endorsement. Understand termination rights in long-running relationships.

For platforms and distributors, streamline notice processing and keep records that demonstrate compliant responses. Ensure takedown workflows incorporate fair use consideration. Be careful about product messaging and internal communications when inducement theories are a foreseeable risk.

For brands and advertisers, if you are using a mark as a source identifier, assume ordinary trademark analysis applies and do not rely on an expressive-work shield. Get explicit clearance for synthetic voice and lookalike creative. Align influencer campaigns with disclosure expectations that have real enforcement behind them.

The central practical reality of entertainment law in this period is that the informal assumptions the industry operated on for years — that transformation defends most creative borrowing, that platform scale confers safe harbor, that parody protects brand-adjacent products, that celebrity voice and likeness require literal copying to create liability — have all been tested and found unreliable. The deals, workflows, and clearance processes that were built around those assumptions need to be updated around what the law actually says now.