YouTube has introduced an AI-powered feature that allows users to generate personalized home feeds by typing prompts describing the videos they want [1, 2, 3].
This update shifts the power of content discovery from a passive algorithm to active user intent. By allowing viewers to define their own viewing experience, the platform aims to reduce the friction of finding niche content and provide creators with new avenues to reach specific audiences [1, 2].
The feature first became available to users in the U.S. in May 2026 [1, 4]. YouTube plans to expand the rollout globally later in 2026 [1, 4].
To use the tool, viewers enter a descriptive prompt into a dedicated box. The AI then curates a home-feed tailored to those specific keywords, or themes [1, 2, 3]. This represents a move toward more granular control over the user interface, moving away from a one-size-fits-all recommendation system [1, 2].
While the official rollout began this month, some reports indicate that YouTube may have been testing versions of a more controllable home page as early as November 2025 [5]. However, the current AI-prompted system is the first public iteration of this specific technology [2].
The company said the goal is to improve how users discover new videos and give them more agency over the media they consume [1, 2].
“YouTube launched an AI‑powered feature that lets users type a prompt describing the kind of videos they want.”
This move signals a broader industry shift toward 'generative UI,' where the interface adapts in real-time to user requests rather than relying solely on historical data. By implementing prompt-based discovery, YouTube is attempting to solve the 'filter bubble' problem, allowing users to intentionally break their existing algorithmic patterns to explore new topics without permanently altering their long-term recommendations.
