
Internet Blocking Methods in Russia in 2026
18 min readUpdated
In Russia, access to internet resources may be restricted by domain, address, connection characteristics, or a set of permitted destinations. The legal basis, technical enforcement, and communications monitoring are different parts of the system. A website or VPN failure alone does not prove blocking.
Sources checked on October 30, 2026. This is an overview of legal provisions, technical mechanisms, and published cases, not our own testing of ISPs or circumvention. This material is informational and is not legal advice.
Two levels: what is banned and how it is enforced
Distinguish the basis for a restriction from how it is enforced.
- Legal level: The applicable provision, a court or authorized body's decision, and the prescribed procedure.
- Technical level: Rules for handling connections on an operator's network or at the service itself.
An app store's distribution decision and a payment refusal need separate consideration. A “not available in your region” message does not identify who imposed the restriction.
| Metric | Result | max | avg | min |
|---|---|---|---|---|
| Ping | —ms | — | — | — |
| Download | —Mbps | — | — | — |
| Upload | —Mbps | — | — | — |
Legal base: Federal Law 149-FZ and the registry
Article 15.1 of Federal Law 149-FZ describes the unified register of domain names, page identifiers, and network addresses of resources containing prohibited information. Inclusion and access restrictions follow the grounds and procedure prescribed by law.
The registry is an example of a blacklist model. This does not mean that every unavailable resource has an entry: the law provides other procedures, and a failure may have a technical cause.
For how blocking works through DNS, IP, and DPI, see How Website Blocking Works: From DNS to DPI.
ORI and data storage
Article 10.1 of Federal Law 149-FZ defines an organizer of information dissemination, or ORI by the operation of systems and software for exchanging electronic messages. It sets obligations to store and provide information in the prescribed circumstances. It also sets additional requirements for instant messaging services.
Calling a service a “social network” or “email” does not replace checking the applicable provision. Data storage and restricting access to a resource are different obligations. For context, see the 2026 overview of Russian data legislation.
TSPU: technical enforcement at the ISP
The obligation to ensure the installation of TSPU — technical means of countering threats — on an internet access operator's network was introduced by Federal Law 90-FZ of May 1, 2019. Its main provisions entered into force on November 1, 2019; some provisions took effect later. This is the legal basis of the so-called sovereign internet law, rather than the “Yarovaya package.”
TSPU is equipment and an enforcement system; DPI is a method of traffic analysis. Article 65.1 of the communications law provides for centralized management of TSPU and binding instructions to network participants. A difference between operators or testing times does not prove who applied which rule.
| Method | What it affects | What one failure does not establish |
|---|---|---|
| DNS | Obtaining an answer about a name | A name resolution error does not prove tampering |
| IP | Access to a network address | A timeout does not identify the cause |
| DPI / SNI | Flow characteristics; a name visible to the filter | A dropped connection does not prove VPN identification |
| Combined | Several rules | Instability does not reveal which rules apply |
DPI can use visible characteristics of an encrypted flow. This does not mean that it reads all HTTPS content. SNI can serve as a signal when the relevant name is visible to the observer. RFC 9505 is an informational document, not a guarantee of available circumvention.
SORM: monitoring and interception, not blocking
SORM supports operational investigative activities on communications infrastructure. Article 64 of Federal Law 126-FZ separately regulates the storage and provision of information and requirements for communications equipment used for activities prescribed by law.
A SORM generation number cannot be treated as a scale from “metadata only” to “all content.” A legal obligation and the actual capabilities of specific equipment are different facts. The presence of SORM alone does not explain a website failure or prove that every connection can be decrypted.
A VPN protects the path to its server for a connection routed through it. The access operator can observe the connection to that server and its parameters; some network visibility shifts to the VPN provider. Correct HTTPS separately protects the content of its own connection. These limits are discussed in EFF's overview of VPN capabilities.
This does not make a user anonymous or remove information in an account, on a device, or held by a recipient. The use of a VPN alone does not establish which data a particular party has retained. See what your VPN provider sees and whether your ISP can see browsing history.
Whitelists: from model to pilots
A whitelist permits selected destinations and denies the rest by default. The scope of the rule needs to be specified: a device, a particular network, or a type of connection.
On September 5, 2025, Russia's Ministry of Digital Development announced a pilot allowing access to selected Russian services during mobile internet restrictions. This is a dated statement about that situation, not evidence that every service is currently available on every network.
A different phenomenon is a carrier offer called a “white VPN.” On June 9, 2026, Beeline's CEO told RBC that its bee subscription provided access to Spotify, Netflix, Ticketmaster, and Brawl Stars. This is a company statement, not our test or a description of a universal tunnel. See the separate article on Beeline white VPN limits and alternatives.
| Model | Logic | Limit of the claim |
|---|---|---|
| Blacklist | Listed destinations are restricted | One filtering model |
| Carrier “white VPN” | Claimed access to selected services | Check the offer's terms |
| Full whitelist within a given network | Only listed destinations are permitted | Establish the scope of the rule |
The models and dated examples are distinguished at the start of the whitelist section.
What is a whitelist
A whitelist is a filtering model where access is granted only to pre-approved resources, while everything else is blocked by default.
This is the opposite of a blacklist:
- Blacklist: only specific websites are blocked.
- Whitelist: everything is blocked except approved websites.
In this model, the internet is no longer an open environment but becomes a restricted access system.
How whitelists work in practice
An allow rule may apply on an operator's network, during DNS or connection processing, or on a device. These are different points of control.
The simplified logic is:
- The system matches the request or connection characteristics available to it against the rule.
- A matching permitted destination is allowed within the scope of that rule.
- Other destinations are restricted by default.
Allowing the main domain does not guarantee that the entire app works: it may need other addresses, APIs, and services. Receiving a DNS answer alone does not mean that a connection is available. These models and their technical limits are discussed in the informational document RFC 7754.
Risks and consequences
Adopting a whitelist model introduces several serious risks:
Restricted access to information
If the rule covers the entire connection, sources outside the permitted set become unavailable through it.
Centralized control
Whoever maintains and applies the list determines the available set; the procedure for changes and the opportunity to challenge decisions matter.
Reduced innovation and competition
A new service may need its addresses and dependencies added; the result depends on the admission procedure.
Errors and overblocking
A missing address or an incorrect rule may disrupt a permitted service.
Illusion of security
Inclusion on the list does not certify the safety of an app, account, or content. Malicious behavior is possible through a permitted resource too.
Technical feasibility and bypassing
The options depend on which connections the network permits at all. VPNs, proxies, Tor, and mirrors also need an available route. Changing DNS concerns name resolution and does not override a rule based on an address, protocol, or permitted set of connections.
When external connectivity is disconnected, these tools do not create it by themselves. A method's name cannot promise access on a particular network.
VPN and circumvention tool blocking
A VPN may be unavailable because of restrictions on an address, port, or connection characteristics. Obtaining an app and paying for a service are separate conditions; their failure does not prove that the tunnel is blocked.
Ordinary services using TLS do not guarantee VPN availability. A network can permit a limited set of connections or disconnect external connectivity. VLESS, obfuscation, and bridges belonging to a particular tool are not universal guarantees of passing a filter.
See whether VPNs can be blocked completely, how to bypass VPN blocking, and the VLESS guide.
A VPN malfunction and the legal assessment of a user's actions are different questions. Article 15.8 of Federal Law 149-FZ regulates measures concerning access tools; it does not authorize every action taken through them.
Article 13.53 of the Code of Administrative Offences provides liability for intentionally searching for materials known to be extremist, as defined in that provision, and obtaining access to them, including through circumvention tools. This is not equivalent to any accidentally opened page or merely having a VPN installed. A particular situation requires a separate legal assessment. For context, see Can You Use VPN in Russia.
Off-network restrictions: stores, payments, platforms
Distinguish the following:
- App availability in a store: Check the region, version, and the store's own message.
- Payment: A working website does not establish the requirements of a particular seller or payment method.
- Regional terms and accounts: A service may apply its own access and registration rules.
- Hosting and CDN operation: A fault or termination of service can make a resource unavailable without a filter on the user's network.
Changing the route does not replace meeting account or payment requirements. No universal current list of working cards and stores has been established here.
How measures stack in 2026: overview
| Part of the system | Role |
|---|---|
| Legal basis | A provision and the prescribed restriction procedure |
| Network enforcement | Operator rules, TSPU, and other means of handling connections |
| SORM | Supporting activities prescribed by law, separate from the cause of a website failure |
| List of available services | A permitted set within a specified restriction scope |
| Carrier offer | Access to services under published service terms |
| Stores, payments, and platforms | Their own distribution, payment, and account restrictions |
This maps different parts of the system; it does not claim that every failure passes through one registry and one device.
Can the internet in Russia be fully restricted
The completeness of a restriction needs to be defined for a particular user, network, type of connection, and time. Unavailable mobile internet in one region is not the same as disconnecting every network in the country.
What full internet blocking in Russia usually means
Distinguish three scenarios:
- No external connectivity: An external resource is unavailable through this connection.
- A limited set of destinations is permitted: Some services work; the rule does not allow others.
- Selective blocking or throttling: Some connections are unavailable or degraded.
A few working websites do not disprove a serious restriction, but one failure does not prove a complete disconnection either. See the related discussion of why blocking is not about security.
Can Russia technically isolate the Runet
The external network can become completely unavailable through a particular connection. This does not require identifying every possible tunnel: not providing external connectivity or allowing only selected connections is enough.
The feasibility and extent of isolating an entire country are broader questions. One app, operator, or surviving route cannot establish them. There is no confirmed forecast of the Runet's ultimate configuration or guarantee of circumvention here. Network disconnection as a distinct method is described in RFC 9505.
Broad blocking and “gray” throttling
Throttling differs from a complete failure: a connection may remain while data transfer degrades. In Censored Planet's research on Twitter in Russia in March 2021, the authors found an SNI-based trigger and transfer rates limited to approximately 130–150 kbps in the studied connections.
This is another team's historical measurement, not our test or a benchmark for 2026. One slow load does not justify attributing the same rule to a current connection.
DPI-driven controlled access
DPI has limitations:
- Encrypted content and visible connection characteristics are different data.
- The result depends on the rule, available characteristics, and filter implementation; a technology's name does not prove error-free identification.
- An overly broad rule may affect other connections and service dependencies.
These are limits of the method, not proof that access cannot be closed to a particular user.
Whitelisting (only approved services)
When only selected destinations are permitted, the service's dependencies matter: APIs, sign-in, content delivery, and updates may use different addresses. If some required connections are not permitted, an app may work only partially or fail.
The consequences depend on the scope of the rule and the completeness of the list. Without data about a particular system, the number of affected services cannot be established, nor can this scenario be called the most destructive.
Restricting external routes
At the routing layer, it is possible to constrain external directions, reduce international capacity, or change routing priorities. For users, this can look like “the internet is on, but nothing loads”: connectivity exists, yet routes break, latency spikes, and a portion of resources vanishes.
What happens in Russia if restrictions become near-total
Business. If a process depends on unavailable payments, email, cloud services, logistics, or APIs, it may stop or require a replacement. The size of any loss depends on the particular dependencies.
Cybersecurity. Unavailable updates, monitoring tools, and official software sources may hinder protection. This does not prove infection or make an unofficial build an acceptable replacement. Baseline risks are discussed in internet threats and protection.
Education and professional work. Restricting repositories, documentation, papers, courses, and communities may disrupt learning and collaboration.
Information sources. Sources excluded from the available set become harder to access through that connection; their effect on an audience requires separate research.
Everyday services. Banking, messaging, updates, or support may depend on unavailable components. It cannot be assumed that every such service will necessarily stop working.
Why blocking is not about security
Restricting access to information and protecting against a specific network threat are different goals. The explanation “for security” alone does not show whether a measure is necessary and proportionate. Assessment requires its grounds, selection criteria, result, and side effects.
Where filtering helps and where it stops
Filtering can restrict access to a known malicious address. For example, Google Safe Browsing warns about phishing pages and dangerous downloads. This is a specific protective function, not proof of the benefit of every government restriction.
Blocking an address does not remove a malicious file from a device or recover a stolen account. New addresses and other distribution channels require separate detection.
The illusion of security
The absence of a warning or block does not certify that a resource is safe. A list may not cover a new threat, and a permitted app may request unnecessary access.
Keep the usual account and device protections. Do not bypass a certificate warning merely because the website opens through another route.
What determines the technical result
The result depends on where the rule is applied and what it does. A VPN, proxy, Tor, or mirror may be unavailable on the same network. Encrypted DNS protects its own exchange, while HTTPS protects the content of the corresponding connection; they do not override every address or route restriction.
One user's ability to restore access does not measure a restriction's result for the entire audience. No circumvention within minutes is promised here.
Side effects of blocking
An overly broad rule may affect:
- Other resources on a shared address or CDN infrastructure
- Connections to a permitted service
- Work processes that depend on an unavailable resource
- APIs, component downloads, and updates
The occurrence and extent of such effects need separate checks. A timeout alone does not establish a side effect of filtering.
Blocking as a tool of control
Selective restrictions on sources give whoever sets the rules the ability to influence the available information set. This can be used for censorship and pressure on resources.
However, the technical mechanism alone does not establish the motive for a particular decision. Check the published grounds, affected resources, and consequences. The relative frequency of motives has not been measured in this article.
How to assess grounds and proportionality
To assess a restriction, compare its stated goal with its scope. Which resources are affected? Are the grounds and criteria published? Can the decision be challenged? Is information available about results and collateral damage?
Without such information, a security claim cannot be accepted as proof of proportionality. Nor can one inevitable sequence of expanding restrictions be attributed to every case.
What users see and what to do
| Situation | What is known | What to clarify |
|---|---|---|
| An operator's blocking page | There is a restriction message | The grounds and resource address; one message does not identify the mechanism |
| A website works through a VPN | The routes produced different results | The cause of the difference and the service's rules |
| VPN connected, but nothing loads | The client reports a connection | The app route, DNS, server, and exact error |
| An app is absent from the store | It is unavailable in this store view | Region, compatibility, and the official source |
| A region error | The service reports a restriction | Product and account terms |
| Only selected services work on the network | Selective availability is observed | The announced mode, list, and restriction scope |
One situation does not establish the cause or guarantee suitable circumvention. Do not disable mandatory protection or install an unknown client for a test. The scope of complete unavailability is discussed separately.
For detailed diagnosis, see how to check that your VPN works.
How to distinguish causes of unavailability
Start with the exact refusal message and the scope of the problem: one website, an app, a mobile connection, or the entire network. Compare it with an official service or operator announcement. One symptom cannot establish a filtering rule, the presence of monitoring, or the legal status of an action.
If the cause is unclear, send official support a description without profiles, keys, subscription links, or raw logs. On a work device, agree on further actions with the administrator.
VPNON is our service. If you decide to test it for your task, read the trial terms before paying: the final price, renewal, next charge, and cancellation. Cancellation and a refund are different actions; read the refund terms separately. A purchase does not guarantee access on every network.
Author: Adam Jensen — Cybersecurity specialist, VPNON