Social media platforms are shifting from centralized recommendation systems to user-controlled algorithms, enabling users to personalize their feeds. This change reflects a broader trend toward more flexible and customizable content discovery, similar to streaming services. Platforms like Threads, Instagram, and TikTok are introducing tools that let users influence what appears in their feeds by expressing preferences directly to the algorithm. This move aims to enhance user engagement by displaying content that aligns with individual interests. The shift marks a significant evolution in how recommendation systems operate, moving away from a one-size-fits-all approach to a more personalized experience.
Threads introduced a new 'Your Algo' feature on June 16, 2026, building on its earlier 'Dear Algo' tool. This feature allows users to set private preferences for content visibility, specifying topics they want to see more or less of, with options for one, three, or seven-day durations. For example, a user could request to see more baseball content and less stressful news. Instagram also launched a tool called 'Your Algorithm' in early June, which lets users view and adjust the topics shaping their recommendations across the feed, explore, and reels. The tool was initially available for the Reels feed in December 2025 and now extends to all content areas.
Instagram's head, Adam Mosseri, noted that traditional recommendation models lacked transparency, but large language models (LLMs) now allow for more understandable systems. He explained that these models can show users why content is displayed and let them communicate their preferences explicitly. TikTok's 'Manage Topics' tool, launched in 2024, lets users customize their 'For You' feed by adjusting sliders for different topics. In 2025, TikTok expanded this with AI-powered Smart Keyword Filters that automatically limit content containing related keywords, such as synonyms. For instance, filtering out 'remodeling' would also block 'renovation' and 'renovations'.
Source: techcrunch
