Threads, Instagram, and TikTok are all rolling out tools that let you train your own recommendation algorithm, and the latest of them went live this week. According to TechCrunch AI, the three platforms are now letting people directly tune what shows up in their feeds, moving beyond the old “Not Interested” tap. TechCrunch AI reports this marks a real shift in how recommendation systems work: feeds are turning from a one-size-fits-all TV channel into something closer to a streaming service you can dial in.
What stands out here is who’s giving up control. For years the algorithm decided, and you reacted. Now the platforms are handing some of that power back, partly because large language models make it possible to explain why content appears and let users say what they actually want.
What each platform launched
Threads introduced “Your Algo” on June 16, 2026. It builds on the older “Dear Algo” tool from February, which let you steer your feed by posting publicly (think “Dear Algo, show me more posts about podcasts”). The new version keeps it private:
- Tell Threads you want more or less of specific topics, no public post required.
- Set how long the request lasts: one, three, or seven days.
- Example use: more baseball, less stressful news, for a week.
Instagram rolled out “Your Algorithm” in early June. It first appeared on the Reels feed back in December 2025, and now it covers feed, explore, and reels together:
- See the topics Instagram thinks you care about most.
- Tell the app what you want more or less of, and recommendations adapt.
- Find it in your settings.
Instagram head Adam Mosseri framed the thinking behind it. He’s said ranking models were historically built with technology that wasn’t transparent to users, but LLMs can now make those systems more understandable by showing why content is displayed and letting people state their preferences outright.
TikTok has offered “Manage Topics” since 2024, and it’s the most granular of the three:
- Adjust a slider per topic (sports, travel, humor, current affairs, dance, food) to see more or less.
- Tap an info button to check what a category covers. TikTok says “Creative arts” includes painting, drawing, graphic design, and art tutorials.
- In 2025, TikTok added AI-powered Smart Keyword Filters that catch synonyms. Filter “remodeling” and it also blocks “renovation” and “renovations.”
How they compare
Each platform solves a slightly different piece of the same problem. Threads is best for temporary tuning, since you can set an expiry and let the feed snap back. Instagram leans into transparency, showing you the topic profile it built about you before you edit it. TikTok gives the finest control with sliders and keyword filtering, which is useful if you want to actively suppress a subject rather than just see less of it.
Why it matters
This is significant because it changes the deal between users and platforms. A tailored feed is the obvious win for you. But there’s a business reason the giants are doing this, and it’s worth saying plainly: customizable algorithms boost engagement by surfacing the content people are most likely to consume. More control for you, more time-on-app for them. Those goals happen to line up here.
The LLM angle is the quiet story underneath all three launches. Recommendation systems used to be black boxes even the companies struggled to explain. Putting language models in the loop makes it realistic to show your topic profile and act on plain-language requests like “less stressful news.” That’s a different relationship with the feed than a thumbs-down button ever offered.
A few caveats are worth keeping in mind. These tools shape recommendations, they don’t hand over the full ranking system. Threads requests expire on a timer, so tuning isn’t permanent unless you keep at it. And how much any of this actually reshapes your feed in practice will vary by how aggressively each platform weights your stated preferences against everything else it tracks.
The direction is clear. Feeds are becoming something you configure, not just something served to you. Expect more platforms to follow, and expect the controls to get more detailed. For the full breakdown of each tool, check the original report at TechCrunch AI.