At the end of March 2026, Telegram released a major update that significantly expanded poll capabilities: adding media to options, user suggestions for new options in an active poll, disabling revoting, shuffling options for each user, time limits, hiding results until completion, viewing results by the creator without voting, and other settings.
Against the backdrop of this update, Pavel Durov separately noted that the team is «already working on the next step» — and directly named directions related to anti-rigging:
- poll statistics;
- voting restrictions for channel subscribers only;
- voting restrictions by user countries.
The key point in the message: rigging and «rigging» of votes by bots in large channels has become a large-scale problem, and Telegram is building tools to «stop» it.
Why the topic of rigging has become public and urgent
In parallel with the announcement of «next steps», Durov became involved in the resonance around a political poll in Telegram: in a comment to a poll about elections to the State Duma he wrote that «the overwhelming majority» of votes for United Russia (more than 90%) in this vote were added by bots from «exotic countries»; it was also reported that the bots were blocked, and work on additional functions against bots in polls is underway.
An important technical detail (which helps understand the logic of future country-based restrictions): in the discussion of this situation, the author of the channel where the poll took place explained that it was not about foreign IP addresses, but about accounts registered to SIM cards from «exotic» countries — due to the lower price of such SIMs during mass purchases for bot networks.
It is also worth noting that Telegram has already added measures that «complicate targeted rigging» — for example, shuffling answer options «for each user». This function is directly named among the new poll capabilities (as part of the March update). However, according to the observations of the author of the aforementioned poll, such measures alone may not be enough: he recorded that the «flood» of votes continued, and directly asked for the option to vote «only for subscribers».
What tools may appear and how they are useful
Below are the three announced directions and the practical meaning of each (with the caveat: Telegram has not yet published the exact UX/format and criteria, so part of this is analysis by analogy with existing mechanics).
Poll statistics
The idea here is clear: if an administrator sees not only final percentages but also dynamics (time, spikes, distribution by audience segments), it is easier to distinguish organic «growth» from anomalies. Telegram is already moving in this direction: for polls with visible votes, timestamps have been introduced showing the time of each vote — this significantly increases transparency in «non-anonymous» scenarios.
Voting only for channel subscribers
This is one of the most anticipated barriers against «raids» by external audiences and cheap bot packs that are not part of the community. Durov directly names this as a planned option.
An important nuance for real effectiveness is the question of «subscription tenure»: if any subscribers «right now» can vote, bots will be able to mass subscribe and vote. Therefore, in practice, the market usually needs options like «only subscribers at the start», «only subscribers older than N days», etc. (this is not in the announcement yet, so it remains an open question).
Restriction by user countries
The idea correlates with the case of «exotic countries» described above: if rigging is built on cheap SIM profiles from certain regions, filtering by country sharply increases the cost of an attack.
Telegram already has a similar pattern in another tool — giveaways in channels: there the organizer can restrict participants to «residents of certain countries» (and/or only new subscribers), meaning the audience segmentation mechanic already exists.
There is also a historical precedent for using «geo-restrictions» in polls: according to the Telegram community, in 2020 the Telegram service account sent a poll to users from Belarus where only users with Belarusian SIM cards could choose a candidate, and others could choose the option «I am not from Belarus».
How this will likely work technically
Although Telegram has not yet shown the «settings fields» for future anti-rigging options, existing tools provide sufficient hints.
First, in giveaways Telegram directly allows restrictions by country and by audience subscription type. This means that on the server side there already exists: (a) a way to determine a user's «country affiliation», (b) a way to verify the fact of subscription/compliance with conditions. Moreover, in the descriptions of giveaway updates, it was separately emphasized that some of the new features are enabled «on the server side».
Second, the Telegram community describes that the «country» binding in such restrictions often depends on the phone number linked to the account (i.e., the country of the SIM/number), rather than the current IP. This was directly mentioned in the analysis of the giveaway restriction mechanics. This approach logically aligns with the story about bots on «cheap SIMs from exotic countries».
Third, Telegram is simultaneously strengthening «observability» of polls: in addition to timestamps of votes in visible polls, the March update added a whole block of settings that make interference more noticeable or expensive (for example, disabling revoting, hiding results until the end, time limits, showing who voted). In practice, this means that «poll statistics» as a separate function looks like a logical continuation of the started line.
What this changes for channel owners and advertisers
The main effect of the future tools is increased trust in polls as a «feedback tool», especially in large channels where rigging is cheaper and more noticeable at scale. Telegram's logic here is common with the logic of giveaways: the company directly positions its built-in mechanics as a way to ensure a «fair result» (in giveaways — a fair choice of winners from those who meet the conditions).
For channel administrators, this may provide practical advantages in three scenarios.
The first scenario is editorial and product polls (what to publish, which sections resonate, which monetization formats do not irritate the audience). Here, statistics and the ability to limit «external noise», if added by non-subscribers, will be especially useful.
The second scenario is polls as part of an advertising campaign (a brand wants «social proof» through voting). Here, new restrictions may reduce demand for «rigged cases» and make quality placements more valuable.
The third scenario is political/social polls, where the motivation for manipulation is higher. The public example with Durov's comment about «>90%» bots shows that Telegram is aware of reputational risks precisely in such cases.
What is still unknown and what risks remain
Today Telegram has confirmed the development direction, but has not provided specifications: there are no release dates, no description of the «poll statistics» interface, no answers to questions like «will there be a filter by subscription tenure» or «can multiple countries/regions be restricted».
There are also inevitable compromises.
The «subscribers only» restriction will reduce external raids, but will not solve the problem of «subscribed bot farms» if a «subscription age» criterion or additional anti-fraud signals do not appear. This is clearly visible from discussions around rigging: even with complicated access and bot cleanups, players remain who adapt through proxies and changes in GEO patterns.
Country restrictions can improve data quality, but at the same time cut off part of the real audience (diaspora, relocators, people with a foreign number while actually residing in the «target» country). At the same time, if «country» is determined by phone number, users with virtual/corporate numbers may fall into the «wrong» segment.
Finally, «poll statistics» itself is a tool that helps identify anomalies, but is not an automatic «honesty button»: rigging can become more careful (stretched over time, mixed across segments), which means admins will still need data interpretation and sound control of methodology (for example, separating polls «about subscriber opinion» and polls «about reach outside the channel»).
