<\/p>\n
Google is testing a more granular form of personalisation for its Discover feed that allows signed-in users to tell the service, in plain language, what they want to see and what they do not.<\/p>\n
The test signals a shift toward explicit, AI-driven control of recommendation systems, a change that could alter how audiences encounter news and how publishers compete for attention on one of the web\u2019s most important distribution surfaces.<\/p>\n
Announced through Google\u2019s Search Labs experiments and reported by several outlets, the feature extends a broader industry move toward promptable recommendation tools already appearing on platforms including Spotify, Threads, Instagram and YouTube.<\/p>\n
According to reports by 9to5Google<\/a> and Android Central<\/a>, the feature appears as \u201cTailor your feed\u201d inside Search Labs. It opens a prompt box where users can type natural-language requests such as \u201cKeep me updated on the Bundesliga\u201d or \u201cShow me posts from OnlineMarketing.de and t3n.\u201d Discover\u2019s AI then updates the feed in real time and retains those preferences.<\/p>\n \u201cDiscover\u2019s AI will update your feed instantly and remember your preferences, which you can adjust anytime,\u201d Google said.<\/p>\n Follow-up prompts and an AI chat thread allow users to refine tone, formats, publishers, creators or exclusions, for example asking only for vegan or gluten-free recipes. Android Authority and Business Standard report the experiment is currently limited to Search Labs participants in the United States and to US English.<\/p>\n The test builds on earlier Discover experiments. Android Central<\/a> and TechCrunch<\/a> note Google has previously trialled features such as Daily Listen, an automatically generated audio briefing drawn from a user\u2019s Discover feed, alongside changes that introduce social posts and short-form video.<\/p>\n In a blog post outlining Discover updates in September 2025, Google said it was also working to surface more content types, let users follow publishers and creators directly and offer previews before following.<\/p>\n For users, the appeal is straightforward. Prompt-based controls promise faster and more precise tailoring of recommendations without repeatedly training the algorithm through clicks or \u201cNot interested\u201d signals.<\/p>\n Observers also see the feature as part of a wider push toward AI-assisted personalisation that can support specific projects, moods or formats, and in some cases reduce the need to visit external sites for summaries or audio briefings.<\/p>\n The same reports highlight risks for publishers and for public discourse. More explicit preference-setting tightens the connection between declared interests and advertising targeting, improving ad relevance for platforms. Business Standard and 9to5Google warn that highly granular feeds could deepen content bubbles, limit serendipitous discovery and make it harder for smaller outlets to reach new audiences.<\/p>\n The ability to prioritise or exclude specific sources may concentrate attention on already dominant publishers and creators.<\/p>\n