YouTube Lets Users Prompt AI for Custom Homepage Feeds

YouTube is rolling out a new artificial intelligence tool that allows users to generate personalized video feeds through text prompts, marking the latest effort by the Google-owned platform to refine its recommendation algorithms. The feature, called "Your custom feed," enables signed-in viewers in the United States to describe desired content such as short meditations or in-depth podcasts, after which the system assembles a tailored selection of videos that can be pinned to the homepage.

The rollout, announced this week, is limited to English-language users on mobile apps and desktop browsers. To activate the option, individuals must navigate to a tab labeled "Your custom feed" and enter a description of their viewing preferences. Prompts can be adjusted later to produce different results, and users retain the ability to flag unsatisfactory outputs through a feedback menu. Access depends on having search and watch history enabled in account settings, a requirement that ties the tool directly to YouTube's existing data collection practices.

This development follows similar initiatives at other major platforms. Spotify introduced prompted playlist generation earlier this year for music and podcasts, while Instagram expanded topic-based controls for its Reels algorithm last December. YouTube's version stands out for its emphasis on open-ended descriptions rather than fixed categories, potentially broadening the range of content it can surface. Company statements highlight flexibility for users seeking content aligned with moods, hobbies, or routines, yet the underlying system still draws from the same vast repository of videos that powers the platform's standard recommendations.

Observers note that the feature arrives as technology companies accelerate deployment of generative AI across content platforms. YouTube has simultaneously worked to label AI-generated videos, reflecting ongoing efforts to manage the technology's proliferation. The new feed tool, however, places greater responsibility on users to supply detailed instructions while relying on historical viewing data to refine outputs. Critics of concentrated corporate power in digital spaces have long raised questions about how such systems prioritize engagement metrics over diverse or independent sources, and this update could intensify those discussions by further embedding algorithmic curation into daily browsing.

The requirement to keep watch history active underscores the trade-offs involved. Without it, the custom feed option remains unavailable, limiting participation to those comfortable with ongoing tracking of their activity. YouTube has indicated that prompts can be revised at any time, but the process still funnels user intent back into the same recommendation engine that determines visibility for creators across the site. As the feature expands beyond its initial United States audience, its impact on content discovery and data practices will likely draw continued attention from regulators and advocacy groups monitoring big technology's influence over information flows.