Why Fixed VPN and Proxy Endpoints Get Blocked

Every major internet censorship system in the world, from state-run firewalls to targeted regional blackouts, works the same basic way: identify the servers or addresses that circumvention tools rely on, then block them. According to a new 2026 research paper by Anindya Maiti, this affects an estimated four billion people living under some form of internet censorship today.

The core problem, as the paper describes it, is that deployed circumvention systems share a structural weakness. Their endpoints (the specific IP addresses, domains, or bridge relays that users connect to) are fixed. That means a patient censor with enough time and resources can eventually enumerate every endpoint a tool uses and add them to a blocklist. This is exactly what has happened repeatedly with VPN services and proxy tools in heavily censored countries: a new set of servers goes up, works for a while, and then gets systematically discovered and blocked.

This cat-and-mouse dynamic isn't new, but it has generally favored the censor. Standing up new infrastructure takes time and money for circumvention developers. Blocking a known IP address takes a censor almost no effort at all once it's been identified.

How Endpoint Rotation and Federated Learning Work in Plain Terms

The research proposes a different model called moving-target censorship resistance, nicknamed "Block-A-Mole" after the arcade game where targets pop up unpredictably and disappear before you can hit them. Instead of relying on a fixed set of servers that a censor can eventually map out, the system rotates endpoints so that what worked yesterday may not be the same thing working today.

The twist in this particular approach is the use of federated learning, a machine learning technique where multiple parties collaboratively train a shared model without pooling their raw data in one central place. In the context of censorship circumvention, this kind of distributed, decentralized approach could theoretically let a network of users and relays adapt collectively to blocking patterns, without any single point of failure or a single list of endpoints that a censor could obtain and neutralize all at once.

In plain terms: rather than a small number of predictable doors that a censor can watch and eventually lock, the system aims to create a constantly shifting set of doors, with the knowledge of where those doors are distributed and refreshed across the network rather than concentrated in one place.

What This Means for Activists and Journalists in Censored Countries

For people working under repressive internet controls, journalists, human rights defenders, and ordinary citizens trying to access independent information, the practical value of this kind of research is about sustainability. Circumvention tools that rely on fixed infrastructure tend to have a shelf life. Once a censor's engineers catch up, the tool stops working until developers can stand up new servers, and the cycle repeats.

A system designed around the assumption that endpoints will be found and blocked, and that treats rotation as a built-in feature rather than a reactive patch, could in principle make circumvention tools more durable over time. That matters most in countries where internet shutdowns and selective blocking are used as tools of political control, and where access to unfiltered information can carry real consequences for the people trying to reach it.

It's worth noting this is a research proposal, not a product. The paper describes an approach to the problem, not a deployed service that people can download today. Real-world adoption, if it happens, would likely take years and would need to survive contact with the same adversaries the research is trying to outmaneuver.

Limitations and How Far This Is From Real-World Deployment

Moving-target strategies also have their own known trade-offs. Rotating endpoints too aggressively can create usability problems, some legitimate users may struggle to reconnect if the system changes too unpredictably. Coordinating a federated, decentralized network also introduces its own security questions: how do you verify new endpoints are legitimate and not run by an adversary trying to infiltrate the network? These are exactly the kinds of questions academic research in this space needs to work through before anything resembling this reaches ordinary users.

For now, the paper represents an early but notable contribution to a long-running arms race between censors and the people trying to route around them.

What This Means For You

If you live in or travel to a country with heavy internet restrictions, moving-target censorship circumvention research like this is a signal of where the field may be headed, not a tool you can use right now. The practical reality today is that people still rely on existing VPNs and proxy services, imperfect as they are, to get around blocking. If you're evaluating options for a specific country, resources like our Best VPN for Ethiopia guide walk through what currently works and what limitations to expect in a challenging censorship environment.

It's also a reminder that circumvention technology and the policy debates around it are connected. Decisions like the Lofgren-Tillis Bill show how VPN access and infrastructure are shaped as much by legislation as by technical innovation.

Takeaways

  • Fixed VPN and proxy endpoints remain vulnerable to being enumerated and blocked by determined censors, a structural weakness the research aims to address.
  • Moving-target censorship circumvention using federated learning is a research concept, not a deployed tool, and real-world use is likely years away.
  • People currently facing internet censorship should rely on established, actively maintained circumvention tools and country-specific guidance rather than waiting on emerging research.
  • Keep an eye on how circumvention research evolves, since durability against blocking is one of the biggest unsolved problems in the space.