Social media is moving through a quiet but significant reset. After more than a decade in which recommendation engines largely decided what people saw, major platforms are giving users new tools to steer their feeds with far more precision. The latest wave of features from Threads, Instagram and TikTok suggests a broader industry shift: algorithms are no longer just a black box behind the timeline, but something users can increasingly inspect, adjust and even teach.
That change may sound incremental, but it marks a major philosophical break from the old model of social networking. Instead of relying only on follows, likes and blunt feedback such as “not interested,” platforms are now offering controls that let people shape the topics, tone and frequency of the content they receive. The move blends user demand for more relevance with the platforms’ own incentives to keep attention by showing people what they are most likely to engage with.
In practice, the new controls are being framed as a more transparent, more personalized feed experience. In theory, they also represent a response to years of criticism that recommendation systems were too opaque, too manipulative and too difficult to understand. The newest tools do not eliminate algorithmic curation, but they do make the system feel more adjustable — closer to a streaming service menu than a fixed television schedule.
A broader shift in how social feeds work
For years, the dominant logic of social media was simple: platforms ranked content on users’ behalf, and users could only indirectly influence the outcome. A follow, a like, a comment or a hidden post sent signals into the system, but the machinery itself remained largely hidden.
That architecture is now changing. The latest generation of recommendation tools gives users a clearer view into what the platform believes they like, while also giving them a way to correct the record. Instead of merely reacting to content after it appears, users can tell the platform in advance what they want more of and what they would rather avoid.
That is an important distinction. It means the user is no longer only a passive audience member. They are becoming a co-designer of the feed.
This evolution is also happening at a moment when social apps are competing harder for attention across video, text and discovery surfaces. A better-tuned recommendation engine can improve satisfaction, but it can also increase watch time, make recommendations feel smarter and reduce the friction that comes from irrelevant posts crowding a feed.
Why platforms are embracing user control
There are two obvious reasons for this trend. The first is user trust. Platforms have spent years under pressure for showing too much outrage, too much spam or too much content that users did not ask for. Giving people more control is a way to soften that criticism.
The second is engagement. If a feed feels more aligned with a person’s current interests, they are more likely to stay, scroll and return. Recommendation systems are not being replaced; they are being refined to feel more responsive.
There is also a technical reason the change is arriving now. Large language models and other modern AI systems can make recommendation logic more legible than older ranking systems. That does not mean the models are simple, but it does mean platforms can more easily present topics, explanations and preference tools in language users understand.
Instagram chief Adam Mosseri has argued that older ranking systems were built in ways that were largely invisible to users, while newer AI tools can help make recommendations easier to interpret and easier for people to influence directly.
Threads adds a private way to guide its algorithm
Threads is one of the clearest examples of this new direction. On June 16, 2026, the platform introduced a feature called “Your Algo,” expanding on an earlier tool called “Dear Algo.”
The earlier feature, which arrived in February, allowed people to post public prompts such as “Dear Algo, show me more posts about podcasts” in order to signal the kinds of content they wanted to see more of. That was a clever and playful approach, but it required users to broadcast their preferences publicly.
The new “Your Algo” tool removes that social layer. Instead of posting a preference for the world to see, users can now make the same kinds of requests privately. The feature lets them specify topics they want emphasized or reduced in their feed, and it also lets them choose how long the adjustment should last.
Threads says the preference can be set for one day, three days or seven days, giving users a short-term way to tune what they see without making a permanent change to their overall experience.
How Threads’ time-based tuning changes the user experience
That time limit is notable. It suggests Threads is not just trying to build a static preference profile, but to let users shape their feed according to what is relevant now. Someone might want more baseball during a playoff week, fewer intense political posts during a stressful day, or more travel inspiration while planning a trip.
This sort of temporary control gives the platform a more dynamic feel. It acknowledges that interests shift quickly and that people do not want the same feed every day. It also provides a middle ground between rigid personalization and fully manual curation.
In a broader sense, the feature reflects an important design principle: users often do not want to dismantle the algorithm, only to nudge it in a better direction.
Instagram makes its algorithm visible
Instagram has also pushed deeper into user-facing control, rolling out a tool called “Your Algorithm.” The feature allows users to see the topics that appear to influence their recommendations and then adjust those preferences to better suit what they want to watch and browse.
The company first introduced this functionality for the Reels feed in December 2025. In early June, it expanded the feature so that it now works across the broader app, including feed, Explore and Reels.
That wider rollout matters because it takes the idea beyond short-form video and into the full Instagram experience. It gives users a central place to inspect how the platform thinks about their interests and to alter the signals that feed those systems.
When users open the feature inside settings, they can review the topics Instagram believes are most relevant to them. From there, they can indicate what they want to see more of and what should be dialed back. In effect, the app is turning recommendation tuning into a user-facing routine rather than a hidden backend process.
Why Instagram’s approach matters for transparency
Instagram’s move is especially important because the app has long been a major reference point in the public debate over algorithmic transparency. Feed ranking is powerful, but it can be hard for users to know why one post appears and another does not.
By surfacing topic labels and giving people direct controls, Instagram is making a subtle but meaningful promise: your feed is not fixed, and the system is willing to show its assumptions. Even if the recommendation logic remains complex, the user can now see and shape part of the profile being built about them.
That does not eliminate the tension between personalization and opacity. Users may still not understand the full mechanics of ranking. But the gap between “the algorithm decides” and “I can influence the algorithm” is getting smaller.
TikTok’s topic sliders and keyword filters
TikTok has been moving in the same direction for longer than some of its rivals. Its “Manage Topics” feature, introduced in 2024, gives users a way to influence what surfaces in the For You feed by adjusting the amount of content they want around specific categories.
Users can access the tool in settings and tune categories such as sports, travel, humor, current affairs, dance and food. Instead of forcing an all-or-nothing choice, TikTok uses sliders to let people increase or decrease how much of a topic they want to see.
The app also tries to make the system easier to understand. If users are unsure what falls under a particular label, they can tap an information button for more detail. TikTok explains, for example, that “Creative arts” includes content such as painting, drawing, graphic design and related tutorials.
AI expands TikTok’s filtering power
In 2025, TikTok added another layer: AI-powered Smart Keyword Filters. These filters broaden a user’s preferences by automatically catching related terms and synonyms. If someone filters out “remodeling,” the system can also exclude related mentions like “renovation” and “renovations.”
That change is important because it reflects a more intelligent, less literal form of moderation and feed shaping. Instead of requiring users to anticipate every possible term, the platform uses AI to infer language variations and apply the preference more broadly.
It is also a sign of how recommendation and filtering are converging. The same AI systems that learn what users might enjoy can also help them avoid themes, keywords and subjects they would rather skip.
What this means for the social media business
The growing emphasis on user-controlled algorithms is not just a product feature story. It is a reflection of the economic reality of social platforms. These companies make money when people stay engaged, and engagement is easier to sustain when a feed feels relevant rather than chaotic.
At the same time, platforms must contend with fatigue. Many users have grown frustrated with feeds that feel repetitive, overly promotional or emotionally exhausting. If the algorithm can be tuned, users may feel less trapped by it — and therefore more likely to keep using the app.
There is a strategic advantage here as well. A platform that gives people a sense of agency can position itself as more user-friendly than one that appears to push content at random. That perception matters in a competitive market where attention is scarce and switching costs are low.
Still, the shift raises a question: how much control is enough? Social platforms are not becoming fully user-run editors. The algorithm remains central. What has changed is that the user now has a better chance of shaping the boundaries within which that algorithm operates.
A move from passive consumption to guided discovery
The best way to understand this trend is to compare it with other media products. Traditional broadcast TV offered almost no personalization. Streaming services, by contrast, let viewers pick genres, queue shows and train recommendations through their behavior.
Social media is starting to resemble the latter model. Users are not selecting every piece of content manually, but they are steering the system with more confidence. They can tell the platform what they want, and the feed can respond in near real time.
That does not necessarily make the algorithm less influential. In some ways, it may make it more effective by aligning itself more closely with what the user is already seeking.
Comparing the new controls across platforms
The details vary, but the direction is similar. Threads emphasizes short-term private adjustments. Instagram focuses on visibility and broader feed-level transparency. TikTok gives users topic-based sliders and AI-assisted keyword filtering.
| Platform | Feature | What users can do | Notable detail | Launch timing |
|---|---|---|---|---|
| Threads | Your Algo | Privately ask for more or less of specific topics | Requests can last 1, 3 or 7 days | June 16, 2026 |
| Threads | Dear Algo | Publicly signal content preferences through posts | Uses a public prompt format | February 2026 |
| Your Algorithm | View and edit the topics shaping recommendations | Available across feed, Explore and Reels | Expanded in early June 2026 | |
| Reels version | Control recommendations in Reels | First version of the tool | December 2025 | |
| TikTok | Manage Topics | Adjust how much of each topic appears in For You | Uses sliders and topic explanations | 2024 |
| TikTok | Smart Keyword Filters | Filter content using AI-expanded keyword matching | Catches synonyms and related terms | 2025 |
Why AI is making these tools possible
Artificial intelligence is doing more than powering recommendations behind the scenes. It is now helping explain and shape them in ways that were harder to build with earlier systems.
Historically, feed-ranking models could identify patterns in clicks, views and engagement, but they were difficult to translate into simple user controls. Today’s AI systems can map topics, classify content more flexibly and generate more understandable interfaces for preference setting.
That makes it easier for platforms to let users manage abstract ideas such as “more sports,” “less stressful news” or “fewer renovation videos,” even when the content itself is messy, shifting or context-dependent.
At the same time, AI gives platforms a way to offer control without giving up automation. The system can continue ranking content at scale while incorporating user input more directly than before.
The upside and the limits of personalization
The upside is obvious: a feed that is more relevant, more comfortable and more aligned with current interests. But there are limits. Algorithmic tuning can improve what shows up without necessarily solving the deeper issues of social media design, such as compulsive scrolling, polarizing content or the emotional intensity of online discourse.
There is also the risk of over-personalization. A highly tuned feed may become too narrow, making it harder to encounter new topics or diverse viewpoints. Platforms are therefore balancing relevance with discovery, and user control with serendipity.
That balance will likely define the next phase of social media product design. Platforms want feeds that feel personal, but not so personalized that they become stale or isolated.
What users should expect next
The spread of these features suggests that more platforms will likely follow with similar tools. As users become accustomed to seeing and shaping algorithmic inputs, lack of control may begin to feel outdated.
Future tools may become more conversational, more visual and more precise. Rather than adjusting broad categories, users may eventually be able to explain preferences in natural language, giving platforms more nuanced direction about tone, subject matter and recency.
That possibility fits with the broader evolution of AI-assisted interfaces. As systems become better at understanding language and intent, the line between editing a feed and instructing an AI assistant may continue to blur.
For now, though, the trend is already clear. The algorithm is no longer something users merely endure. On Threads, Instagram and TikTok, it is increasingly something they can talk back to.
Key takeaways
- Major social platforms are rolling out tools that let users influence recommendation algorithms more directly.
- Threads’ “Your Algo” adds private, time-limited preference controls to the public “Dear Algo” feature.
- Instagram’s “Your Algorithm” lets users inspect and adjust the topics driving recommendations across the app.
- TikTok’s “Manage Topics” and AI-powered keyword filters give users topic sliders and broader content exclusion tools.
- The shift reflects both user demand for transparency and platform incentives to improve engagement.
The bigger picture
The rise of user-controlled algorithms signals an important stage in social media’s evolution. The first era was defined by social graphs — who you follow and who follows you. The second was defined by opaque ranking systems that used engagement signals to decide what mattered. The emerging era appears to be defined by something in between: algorithmic systems that still curate at scale, but increasingly invite the user into the process.
That is not a small change. It alters the relationship between platform and audience, turning recommendation from a one-way broadcast into a guided collaboration. The feed is still algorithmic, but it is becoming more conversational — and for social media companies, that may be the easiest way to keep users feeling both entertained and in control.
