AI in entertainment
Last reviewed
May 30, 2026
Sources
No citations yet
Review status
Needs citations
Revision
v2 · 3,170 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
May 30, 2026
Sources
No citations yet
Review status
Needs citations
Revision
v2 · 3,170 words
Add missing citations, update stale details, or suggest a clearer explanation.
AI in entertainment refers to the use of artificial intelligence across the film, television, music, video game, animation, and live event industries. The technology spans production tools (visual effects, dubbing, voice synthesis, asset and dialogue generation), back-office and creative decision support (script analysis, box office forecasting), distribution (streaming recommendation systems), and consumer-facing creative tools that let anyone generate music, images, and video from text prompts.
Interest in the field grew sharply after 2022, when generative AI systems built on diffusion models and large language models made convincing synthetic media widely accessible. That shift produced both new creative capabilities and significant disputes over labor, consent, likeness, and copyright. Two of the most consequential were the 2023 Hollywood strikes by the Writers Guild of America (WGA) and SAG-AFTRA, and the 2024 to 2025 SAG-AFTRA video game strike, all of which centered substantially on AI, and a wave of copyright lawsuits over the data used to train generative models.
The entertainment industries adopted machine learning gradually. Procedural and rule-based generation appeared in games well before the modern AI wave; algorithmic content has been used to build levels and worlds since the era of titles such as Rogue (1980) and Elite (1984). Streaming services applied collaborative filtering to recommendations in the 2000s and 2010s. In film, studios used software to assist with rotoscoping, compositing, and motion capture cleanup.
Music streaming was an early adopter of data-driven personalization. Spotify acquired the music intelligence company The Echo Nest in 2014, integrating its audio and text analysis into recommendation features.1 In film, AI-assisted de-aging and facial work reached a wide audience with Martin Scorsese's The Irishman (2019), for which Industrial Light & Magic developed a markerless system to digitally de-age Robert De Niro, Al Pacino, and Joe Pesci.2
The inflection point was the broad release of generative tools in 2022 and 2023. Text-to-image systems, then text-to-video systems such as Runway's Gen series and OpenAI's Sora, and music generators such as Suno and Udio, moved synthetic media from research labs into consumer products. The same period saw the first viral AI-music controversies, the 2023 strikes, and the first lawsuits over training data, which together defined the legal and labor landscape of the field.
The table below summarizes how AI is applied across the main entertainment media.
| Medium | Representative uses |
|---|---|
| Film and TV | VFX and compositing, de-aging, voice synthesis and dubbing, localization, script analysis, generative video shots |
| Music | Composition, voice synthesis and cloning, automated mastering, recommendation, fully synthetic tracks |
| Video games | Procedural generation, NPC behavior and dialogue, asset generation, real-time upscaling and frame generation |
| Animation | In-betweening, style transfer, motion synthesis, lip sync |
| Live and immersive | Real-time avatars, generative visuals, interactive installations |
Visual effects and compositing. Machine learning assists with tasks that were once entirely manual: removing wires and markers, generating mattes, cleaning up plates, and synthesizing crowds or environments. Studios have increasingly used neural rendering and generative tools to speed up shots that would otherwise require lengthy 3D modeling.
In 2025, Netflix disclosed what it described as the first use of generative AI to create final footage in a Netflix original, a short shot of a collapsing building in the Argentine series El Eternauta. Co-chief executive Ted Sarandos said on an earnings call that the sequence, produced with Netflix's in-house Eyeline Studios, was completed far faster than a conventional visual effects workflow and that the cost of doing it traditionally "just wouldn't have been feasible" for the show's budget.3 The disclosure was notable because it marked generative video moving into a mainstream, finished production rather than a demo.
De-aging. AI-assisted de-aging has been used to portray actors at younger ages, as in The Irishman and later productions. Specialist vendors and studio VFX houses combine archival footage, performance capture, and machine learning to reconstruct younger appearances.2
Voice synthesis, dubbing, and localization. Voice cloning lets productions recreate or modify performers' voices. The Ukrainian company Respeecher recreated a youthful Luke Skywalker voice for Disney's The Mandalorian and reproduced Darth Vader's voice for Obi-Wan Kenobi, working from archival recordings of the original performers.4 AI dubbing and lip-sync tools, including systems that adjust mouth movements to match a translated track, are used to localize content across languages.
The same techniques have prompted disclosure debates. The 2024 film The Brutalist used Respeecher to refine the Hungarian pronunciation of stars Adrien Brody and Felicity Jones; the film's editor, Davit Jancso (a native Hungarian speaker), fed his own voice into the tool to polish specific vowels and consonants in the actors' existing performances. When the use became public during awards season, it drew criticism focused less on the technology, which the production described as a limited audio touch-up of human performances, than on transparency about how it was used.5
Script analysis and decision support. Studios have experimented with predictive analytics for development and release planning. In 2020, Warner Bros. signed a deal with the company Cinelytic, which markets a machine-learning system that estimates a project's revenue potential based on variables such as cast, genre, and release timing; reporting at the time stressed that the tool was meant to support, not replace, human greenlighting decisions.6 Other vendors, such as Belgium's ScriptBook, have offered automated script analysis and box office forecasting.6
Composition and generation. Consumer tools such as Suno and Udio generate full songs, including vocals and instrumentation, from text prompts. These systems are built on generative models trained on large audio datasets, and they have become central to debates over training data and copyright (see below).
Voice synthesis and cloning. Tools such as ElevenLabs and other voice cloning systems can reproduce a specific singer's or speaker's voice from samples. The most prominent demonstration was the track "Heart on My Sleeve," posted in April 2023 by an anonymous creator using the handle Ghostwriter, which used AI to mimic the voices of Drake and The Weeknd. It went viral on TikTok and accumulated large play counts on streaming services before being pulled from Spotify, Apple Music, YouTube, and other platforms after Universal Music Group, the artists' label, objected. Coverage noted that the takedown was tied in part to the song's reproduction of producer Metro Boomin's signature audio tag, and that UMG had been pressing platforms to act more aggressively against AI tracks trained on its artists.7
Mastering and production. AI mastering services analyze a track's audio characteristics and apply a processing chain automatically. LANDR launched AI-driven mastering in 2014, and iZotope's Ozone offers machine-learning-based mastering suggestions based on a track's sonic profile.8 These tools have made a baseline level of mastering available to independent artists without a dedicated engineer.
Streaming and AI tracks. Fully synthetic acts have tested how platforms handle AI music. The project "The Velvet Sundown" reached more than a million monthly listeners on Spotify in mid-2025 before it was confirmed to be an AI-generated project rather than a human band, raising questions about disclosure and about royalties flowing to tracks with no human performers.9 In response to a broader rise of synthetic uploads, Spotify announced new policies in 2025 that did not ban AI music outright but added measures against spam and impersonation and encouraged AI projects to identify themselves; the company has said it does not require artists to disclose AI use.9
Procedural generation. Games have long used algorithms to generate content at scale. No Man's Sky (2016) procedurally generates its planets, ecosystems, flora, fauna, structures, and ships from seed-based algorithms, producing a vast number of distinct worlds without hand-authoring each one.10 Procedural techniques range from classic noise-based terrain to more recent machine-learning-assisted asset and texture generation.
NPC behavior and dialogue. A newer line of work uses large language models to drive non-player characters that respond with unscripted dialogue. Ubisoft demonstrated a prototype it called NEO NPC, built with Nvidia's Audio2Face animation technology and Inworld AI's language models, in which characters react to players with contextual, real-time dialogue and animation.11 Nvidia markets a related suite, ACE (Avatar Cloud Engine), as a set of "digital human" technologies for game characters that can perceive, reason, and respond in real time; the company lists developers and partners including Ubisoft and Inworld AI.12 These systems remain largely experimental in shipped titles as of 2026.
Asset and dialogue generation. Generative tools are used in development pipelines for concept art, textures, placeholder dialogue, and prototyping, though studio use varies and is often undisclosed.
Upscaling and frame generation. Real-time AI rendering is among the most widely deployed game uses. Nvidia's Deep Learning Super Sampling (DLSS) uses neural networks to reconstruct high-resolution frames from lower-resolution inputs and, in later versions, to generate additional frames. Nvidia describes DLSS 4 as using a vision-transformer model for upscaling and a multi-frame generation feature on its RTX 50-series hardware; in January 2026 the company stated that over 250 games and applications supported multi-frame generation, and at its GTC 2026 conference it previewed a neural-rendering-based successor, DLSS 5, planned for late 2026.13 Competing vendors offer comparable upscaling technologies.
AI assists animation through automated in-betweening (generating frames between key poses), style transfer, motion synthesis from reference video, and automated lip sync that aligns mouth movements to recorded or translated dialogue. These tools are used to accelerate parts of 2D and 3D pipelines, though end-to-end generative animation for finished productions remains limited.
Real-time generative systems power interactive installations, virtual performers, and concert visuals. Real-time avatar and voice systems can drive digital hosts or characters that respond to audiences, and generative visual tools are used to create reactive stage content.
Recommendation is one of the oldest and most consequential uses of machine learning in entertainment, shaping what audiences watch and hear. Streaming recommendation systems combine several techniques.
Spotify's Discover Weekly, for example, has been described as blending three model types: collaborative filtering (finding listeners with similar habits and recommending what they enjoy), natural language processing (scanning text such as articles and blogs to capture how songs are described), and audio models (using convolutional neural networks to analyze a track's tempo, acoustics, and other features so that new or obscure songs can be matched even without listening data).14 Video services use related approaches to surface films and shows. These systems strongly influence consumption patterns and, by extension, which titles and artists gain visibility.
The AI methods used in entertainment overlap with the broader field:
The following are widely cited tools and companies in the field. Capability claims attributed to vendors are described as such and have not all been independently verified.
| Tool or company | Area | Notes |
|---|---|---|
| Runway | Generative video | Gen-series text-to-video models used for VFX and short content |
| Sora (OpenAI) | Generative video | Text-to-video model released to the public after a 2024 preview |
| Suno | Music generation | Consumer tool generating full songs from prompts; named in 2024 RIAA suit |
| Udio | Music generation | Consumer music generator; named in 2024 RIAA suit |
| ElevenLabs | Voice synthesis | Voice cloning and text-to-speech used in media and dubbing |
| Respeecher | Voice synthesis | Used on Star Wars projects and The Brutalist |
| Nvidia | Rendering, NPCs | DLSS upscaling and ACE digital-human technologies |
| Cinelytic | Decision support | Predictive analytics for film development and release planning |
| LANDR / iZotope | Music production | AI-assisted mastering |
AI was a central issue in the 2023 strikes by the WGA (May 2 to September 27, roughly 148 days) and SAG-AFTRA (July 14 to November 9, roughly 118 days), the first time both unions struck simultaneously since 1960.16
The WGA's resulting agreement set rules on AI in writing: AI cannot be credited as a writer, AI-generated material cannot be treated as source or "literary" material that would reduce a writer's credit or pay, and the guild reserved the right to argue that using writers' work to train generative AI violates the agreement or applicable law. The deal also created disclosure requirements around AI use.17
SAG-AFTRA's agreement established consent and compensation rules for digital replicas of performers. Creating an employment-based digital replica requires advance notice, "clear and conspicuous" consent, and payment, with consent and compensation negotiated for each use; the union also secured ongoing meetings with studios about generative AI.18 Some members and observers argued the protections did not go far enough, particularly around how broadly consent could be sought.18
Voice and motion-capture performers, also represented by SAG-AFTRA, struck the video game industry from 2024 into 2025, primarily over AI. The strike lasted about 11 months before members ratified a new Interactive Media Agreement in July 2025 by roughly 95 percent. The deal added consent and disclosure requirements for AI digital replicas and gave performers the ability to suspend consent for generating new material during a strike, alongside pay increases.19
The ability to clone a performer's voice or likeness has raised rights-of-publicity and consent concerns beyond union contracts. The "Heart on My Sleeve" case showed how quickly an unauthorized voice clone can spread and how rights holders respond, while the synthetic-band controversies highlighted disclosure and royalty questions.79 Deepfakes of public figures, including nonconsensual sexual imagery, have prompted legislative and platform responses.
A major unresolved question is whether training generative models on copyrighted works without a license is lawful. In June 2024, the Recording Industry Association of America, on behalf of Sony Music, Universal Music Group, and Warner Music, filed twin copyright lawsuits against Suno (in Massachusetts) and Udio (in New York), alleging the companies copied sound recordings to train their models without authorization.20 The cases evolved into a mix of litigation and licensing: by late 2025, reporting indicated that some major labels had reached settlement or licensing arrangements with the companies while litigation continued for others, with key fair-use questions still pending in court.20 Because terms and the status of individual cases were still developing as of 2026-05-31, specific settlement figures and per-stream royalty rates reported in trade and specialist outlets should be treated as not fully confirmed.
Proponents argue that AI lowers the cost and time of production, makes professional-grade tools available to independent creators, enables localization and accessibility (dubbing, captioning, described content), and supports effects and scales that would otherwise be infeasible. Recommendation systems help audiences find content in vast catalogs.
Critics point to displacement of skilled workers, the use of artists' work as uncompensated training data, erosion of consent over voice and likeness, the spread of deepfakes and impersonation, homogenization of output, and the dilution of royalty pools by mass-produced synthetic content. The strikes and lawsuits reflect an ongoing negotiation over how value, consent, and credit are allocated when generative tools enter creative work.
As of 2026, generative tools are moving from demonstrations into shipped products, as the El Eternauta disclosure illustrated, while the legal framework around training data and likeness remains unsettled and is being shaped case by case and contract by contract. The pace of model improvement, particularly in video and music generation, suggests continued expansion of AI use, alongside continued disputes over consent, compensation, and disclosure.
"Spotify's Discover Weekly explained" and related coverage of Spotify's 2014 acquisition of The Echo Nest. The Sound of AI / Medium. https://medium.com/the-sound-of-ai/spotifys-discover-weekly-explained-breaking-from-your-music-bubble-or-maybe-not-b506da144123 Accessed 2026-05-31. ↩
"How AI is Being Used to De-Age Actors in Movies" and Respeecher coverage of de-aging in The Irishman. Savvy Dispatches; Respeecher. https://www.respeecher.com/blog/de-aging-technology-changing-hollywood-future-film-making Accessed 2026-05-31. ↩ ↩2
"Netflix Says It Used Video-Generating AI for Special Effects in a New Show." Futurism; and "For the first time, Netflix uses GenAI for VFX in original series," Interesting Engineering. https://futurism.com/netflix-video-generating-ai-tv Accessed 2026-05-31. ↩
"How Respeecher's Voice Cloning Brought Young Luke Skywalker to Life in The Mandalorian" and "AI Voice Replication Is Transforming Star Wars." Respeecher. https://www.respeecher.com/case-studies/respeecher-synthesized-younger-luke-skywalkers-voice-disneys-mandalorian Accessed 2026-05-31. ↩
"AI voice technology used in The Brutalist is nothing new; the backlash is about transparency." The Conversation; and Respeecher, "Adrien Brody and The Brutalist: What the AI Controversy Actually Was." https://theconversation.com/ai-voice-technology-used-in-the-brutalist-is-nothing-new-the-backlash-is-about-transparency-248128 Accessed 2026-05-31. ↩
"Warner Bros. To Start Using AI Analysis Tool To Assist In Greenlighting Movies," Unite.AI; and "Cinelytic's A.I. is not green-lighting movies," Fortune. https://www.unite.ai/warner-bros-to-start-using-ai-analysis-tool-to-assist-in-greenlighting-movies/ Accessed 2026-05-31. ↩ ↩2
"Viral AI-powered Drake and The Weeknd song is removed from streaming services," NBC News; and "AI fake Drake track deleted on Spotify, YouTube, TikTok after Universal Music Group copyright claim," Music Business Worldwide. https://www.nbcnews.com/pop-culture/viral-ai-powered-drake-weeknd-song-removed-streaming-services-rcna80098 Accessed 2026-05-31. ↩ ↩2
"The 7 Best AI Tools for Music Production," LANDR Blog; coverage of LANDR (2014) and iZotope Ozone AI mastering. https://blog.landr.com/ai-in-music/ Accessed 2026-05-31. ↩
"The Velvet Sundown explained," Euronews; and "Spotify Embraces AI Music With New Policies, While Combating Spam and Slop," Rolling Stone. https://www.euronews.com/culture/2025/07/08/the-velvet-sundown-explained-whats-behind-the-spotify-verified-ai-band-controversy Accessed 2026-05-31. ↩ ↩2 ↩3
"Procedural generation," No Man's Sky Wiki; and "A Look At How No Man's Sky's Procedural Generation Works," Kotaku. https://kotaku.com/a-look-at-how-no-mans-skys-procedural-generation-works-1787928446 Accessed 2026-05-31. ↩
"How Ubisoft's New Generative AI Prototype Changes the Narrative for NPCs," Ubisoft; and "Ubisoft debuts GenAI-powered Neo NPCs," GamesBeat. https://news.ubisoft.com/en-us/article/5qXdxhshJBXoanFZApdG3L/how-ubisofts-new-generative-ai-prototype-changes-the-narrative-for-npcs Accessed 2026-05-31. ↩
"NVIDIA Digital Human Technologies Bring AI Game Characters To Life," Nvidia GeForce News. https://www.nvidia.com/en-us/geforce/news/nvidia-ace-gdc-gtc-2024-ai-character-game-and-app-demo-videos/ Accessed 2026-05-31. ↩
"Deep Learning Super Sampling," Wikipedia; and "DLSS 4 Technology," Nvidia. https://en.wikipedia.org/wiki/Deep_Learning_Super_Sampling Accessed 2026-05-31. ↩
"Behind Spotify's Discover Weekly Playlist," Cornell INFO 2040 course blog; and "How Spotify Uses AI to Curate Music Playlists," Brainforge. https://blogs.cornell.edu/info2040/2019/10/22/behind-spotifys-discover-weekly-playlist/ Accessed 2026-05-31. ↩
"Diffusion model," overview of diffusion-based generative methods. https://en.wikipedia.org/wiki/Diffusion_model Accessed 2026-05-31. ↩
"2023 Writers Guild of America strike" and "2023 SAG-AFTRA strike," Wikipedia. https://en.wikipedia.org/wiki/2023_Writers_Guild_of_America_strike Accessed 2026-05-31. ↩
"Generative AI in Movies and TV: How the 2023 SAG-AFTRA and WGA Contracts Address Generative AI," Perkins Coie. https://perkinscoie.com/insights/blog/generative-ai-movies-and-tv-how-2023-sag-aftra-and-wga-contracts-address-generative Accessed 2026-05-31. ↩
"SAG-AFTRA Agreement Establishes Important Safeguards for Actors Around AI Use," The Authors Guild; and "SAG-AFTRA's new contract falls short on protections from AI," Prism. https://authorsguild.org/news/sag-aftra-agreement-establishes-important-ai-safeguards/ Accessed 2026-05-31. ↩ ↩2
"SAG-AFTRA Members Approve 2025 Video Game Agreement," SAG-AFTRA; and "Video Game Actors Contract Ratified," Variety. https://www.sagaftra.org/sag-aftra-members-approve-2025-video-game-agreement Accessed 2026-05-31. ↩
"Record Companies Bring Landmark Cases for Responsible AI Against Suno and Udio," RIAA; and "Record Labels Sue AI Music Services Suno and Udio for Copyright," Variety. https://www.riaa.com/record-companies-bring-landmark-cases-for-responsible-ai-againstsuno-and-udio-in-boston-and-new-york-federal-courts-respectively/ Accessed 2026-05-31. ↩ ↩2