Generative AI Shifts Entertainment From Recommendations to Custom Content

Recommendation engines have long steered audiences toward existing content that fits their preferences. Generative AI is changing that dynamic by creating bespoke experiences for each user rather than simply pointing them to pre-made options.

The shift represents a fundamental change in how entertainment platforms approach personalization — moving from curation to creation.

Epic Games' Fortnite rolled out AI-powered conversations for non-player characters to island creators on July 30, exiting experimental testing. The feature replaces scripted dialogue with characters that respond uniquely to each player's actions and words.

Developers define character traits through simple prompts, and the AI generates contextual responses in real time. Instead of hearing identical lines, every player experiences dialogue tailored to their specific interactions.

Kaon AI applies similar principles to narrative video, raising $60 million in Series B funding to build distinct story worlds for each viewer rather than recommending from a fixed catalogue.

If the perfect piece of content for the user was never made, the algorithm cannot help the user.

That's according to Kaon AI CEO Jay Dang, speaking to PYMNTS last month.

Netflix has tested AI-generated advertising creative that visually matches the show a viewer is watching, rather than inserting generic ad breaks across all streams. Brands including DoorDash, Target and TurboTax participated in trials, with Netflix reporting significant improvements in quality and execution.

The streaming service plans to expand the tool to all ad-supported regions by year's end. Netflix is also testing personalized ad frequency caps that adjust which commercials appear based on individual viewing behavior.

Spotify is pursuing a parallel strategy in audio. At the company's 2026 Investor Day, Co-CEO Gustav Söderström outlined a progression from access to personalization to generation.

The streaming platform is developing a Large Taste Model trained on 3.4 trillion daily signals from listener activity across music, podcasts and audiobooks. The system generates individualized audio and remixes rather than only surfacing existing tracks.

Disney's experience illustrates the execution challenges. In December, OpenAI announced a $1 billion equity investment from Disney to let Disney+ subscribers generate fan videos using characters from Disney, Marvel, Pixar and Star Wars through OpenAI's Sora tool.

OpenAI shut down Sora entirely three months later, pivoting toward high-productivity coding and agentic tools. Disney withdrew before any money changed hands.

The original deal included specific restrictions: characters could appear in generated videos, but talent likenesses and voices were excluded. Disney and OpenAI had created a joint steering committee to monitor user-generated content against brand guidelines.

Creating unique content for every viewer carries higher costs than serving pre-existing files. Each generated piece requires compute resources, moderation and error-catching mechanisms at a scale fixed libraries never demanded.

The economics of one-to-one content generation remain unproven at the scale platforms need to serve millions of concurrent users, even as the technology advances.