Aquileo | The Mullvad Blog

Aquileo | Age verification for social media – the beginning of the end for a free internet?

2026-06-01T12:12:40.297708+00:00

So-called age verification for social media is spreading across the world, framed as an effort to create a safer internet for children. In reality, age verification lays the foundation for a fully government controlled internet.

Countries around the world are moving to introduce online age verification. Part of this involves age verification for harmful content (most often pornography, sometimes video games), but above all it focuses on banning social media for children.

The big tech social media companies are bad. Their business model is bad; it is based on mass surveillance and manipulation, and they cooperate with governments in mapping entire populations. But age verification is fundamentally the wrong approach to preventing children from using big tech social media platforms. Introducing age verification is based on the state being able to force social media companies to verify their users’ identities. But the big tech social media platforms already know which of their users are children. Their business model depends on knowing this. They know how old users are, who their friends are and what ice cream they like. As age verification is based on coercion of the social media platforms, politicians could instead force them to stop doing the things politicians consider harmful to children, or force them to block children (again, they know who they are) from using their services. But instead, politicians seek to massively invade everyone’s privacy and undermine democratic rights on a global scale. In other words, the latter is the real objective – they do not want to protect children; they want to impose control.

And impose it they do. Australia has already introduced a social media age restriction for users under 16. The same applies to Indonesia and Brazil. Age restrictions have been approved but not yet implemented in Denmark, Portugal, and Malaysia. In France, an agreement has been reached, though details are still being discussed. Proposals are on the table in Spain and Turkey. In Germany, the major parties agree on introducing age restrictions, and in Sweden the issue is under investigation. The topic is also being discussed in countries such as the Czech Republic, Greece, Austria, Poland, Canada, Slovenia, and the Netherlands. In April 2026, the European Commission launched an EU age verification app, and one month later Ursula von der Leyen presented plans for EU-wide age restrictions. In the United States, half of all states either have pending legislation or have already introduced laws imposing age restrictions for inappropriate content and/or social media. The number of countries preparing age verification measures is growing rapidly. Updates can be followed on Techpolicy.press.

Most age verification is identity verification

As age verification is currently being rolled out, it is up to individual websites and services to implement it as they see fit. As a result, the quality of age verification measures varies greatly. This became clear in the autumn of 2025, when Discord was hacked, exposing the ID documents of 70,000 users. However, there is one common factor in most age verification systems (Zero-Knowledge Proof being an exception, more on that below): if age verification is introduced, everyone will have to identify themselves either to the service/website they want to use or to a third party capable of linking them to their activity on that service/website. The correct term for age verification as it is implemented today is therefore identity verification. Given today’s internet infrastructure, it is unreasonable to assume that this information will not be shared through commercial agreements or with governments.

The consequence of introducing identity verification is therefore that freedom of information is restricted (you can no longer visit regulated websites anonymously) and that you can no longer post anonymously on social media. You cannot be certain that your criticism of the government will not be followed up by the authorities. You can no longer start a digital initiative on a social media platform aimed at gathering people to criticize an authority without facing a significant risk of consequences. Depending on the country you live in, this could even endanger your life. In its current form, social media identity verification removes important tools for activists in countries where criticizing those in power is dangerous.

Freedom of expression is threatened not only in a direct sense (you post something and then the police knock on your door), identity verification also creates a chilling effect. It becomes a cornerstone of censorship machinery in the sense that people begin to self-censor if they know that expressing opinions may have personal consequences. This is also something that changes over time. What is considered acceptable to post online is determined by whoever currently holds power. Different sides of politics often have different views on what constitutes harmful content. Just because what you post today is not considered inappropriate does not mean it will remain acceptable in the future.

Broad and arbitrary legislation, along with mandatory ID tagging for every post, hardly provides a strong foundation for freedom of expression. Today, 30 people are arrested every day in the United Kingdom for posting something online that authorities classify as “grossly offensive.” In Germany, police conduct raids on people’s homes for insulting politicians online. One infamous example is the so-called “Pimmelgate,” where a person was subject to a police search after calling a German politician a term for male genitalia. In the United States, authorities are trying to pressure tech companies into revealing the identities behind accounts protesting ICE. Another example is when Canada introduced emergency acts during the 2022 trucker protests and then used social media to identify demonstrators and freeze the bank accounts of people who financially supported the protest.

The slippery slope of age verification. VPN next?

Restrictions introduced at the national level can be bypassed by changing one’s geographic location digitally, using tools such as VPNs, virtual phone numbers, eSIM cards, Tor and dedicated services. This has already led politicians in several countries to consider introducing identity verification for VPN services (presumably because VPNs are the most common and accessible method of changing digital location).

In the United Kingdom, the House of Lords sent an amendment in early 2026 (regarding the the Children’s Wellbeing and Schools Bill) to the House of Commons, proposing an 18-year age limit for using VPN services. The House of Commons rejected the House of Lords amendment four separate times. However, the House of Commons instead introduced its own proposal, which was passed and has now become law. This agreement grants the government the power to introduce restrictions through secondary legislation, with only limited parliamentary scrutiny. The government has confirmed that it intends to act on this and introduce some form of social media restriction for children under 16. The government has also hinted that it may consider introducing identity verification for VPN usage, effectively joining countries such as China and Russia in opposing VPN services.

The issue has also been raised in France. As Minister for AI and Digital Affairs Anne Le Hénanff put it: “If [this legislation] allows us to protect a very large majority of children, we will continue. And VPNs are the next topic on my list.” Discussions about VPN restrictions have also occurred in the United States. Utah has gone the furthest by introducing a law making it illegal to circumvent restrictions using a VPN. Within the EU, VPN restrictions have been discussed both under the Going Dark initiative and in discussions related to age verification. In response to a direct question about VPNs as a tool for bypassing age verification, EU Commissioner Henna Virkkunen said in April: “Of course, it's an important part of the next steps also to look at that it [age verification] shouldn't be circumvented.”

If VPN services were to implement identity verification, this would mean collecting data that could be abused through either malice or incompetence. It would, for example, make such services risky for whistleblowers and activists, make it harder for journalists to work with sensitive information, and create a chilling effect on online debate (VPNs can help people post anonymously on social media).

If VPN providers were to impose an age limit on their service, this would also mean that underage users would effectively lose their right to online privacy. Ironically, one consequence would be that social media companies mapping people’s lives through third-party trackers on websites could continue monitoring young people’s online behavior via their IP addresses without any interference. In other words, politicians would remove one of the protections children have against the very companies they claim to want to protect children from.

Identity verification in app stores and at the operating system level

As identity verification is now being introduced globally, different parts of the world are implementing it in different ways. Some countries believe the best solution is to impose controls through major app stores such as Apple’s App Store and Google Play. In Australia, Brazil, South Korea, Singapore, and several US states, Apple has already begun introducing identity verification at the App Store level to restrict access to apps containing adult content.

Identity verification in app stores only regulates access to apps. Therefore, several countries have also begun demanding identity verification at the operating system level itself in order to block access to certain websites directly through the OS. In the United Kingdom, Apple has already introduced this – despite there being no law requiring it (Apple is, however, under general pressure from British authorities). Without warning, Apple implemented identity verification on British iPhones through its system update on March 24, 2026. Suddenly, 35 million British users had to identify themselves using either a credit card or a government-issued ID card in order to avoid restrictions on their phones. Users who did not verify their identity saw their devices enter a mode where Apple’s web content filter and communication safety features were automatically activated, limiting which websites could be visited in Safari or any third-party browser, while messaging services and FaceTime were being monitored for inappropriate content.

When Apple introduced OS-level identity verification in the UK, reports quickly emerged of people bypassing it by creating US-based Apple IDs instead. App store-level identity verification can be circumvented in the same way. And this is how things will continue. The situation will not fundamentally change unless Apple and Google (assuming Google also locks down its system and introduces identity verification) implement these controls in their operating systems globally. But even then, there would still be ways to circumvent restrictions.

Users could, for example, turn to open-source operating systems, which by definition cannot be fully controlled because they are open and modifiable. In such systems, no one else can decide which apps you download or which websites you visit, and no identity verification can be imposed unless you choose it yourself. This points to the final stage for countries seeking total control over their citizens’ use of the internet. National identity verification can be bypassed again and again until authoritarian governments are ultimately forced to ban people from owning devices they control themselves. The only question is how this would be implemented. Well, we will soon find out … In Brazil, a law implemented in March 2026 states that identity verification must be carried out both at the app store level and within the operating system itself (including open-source systems), or companies distributing these services will face fines of up to $10 million. A similar law has passed in California which will require identity verification at the operating system level starting in January 2027. Open-source systems were initially included, but later removed, while web browsers and websites were added to the scope. Similar proposals exist in states such as Colorado and New York. In April 2026, a federal proposal was introduced that would require OS-level identity verification across the United States.

The final destination: state-provided phones and computers for all?

It will be interesting to follow the countries that pursue system-level control as they move further down the slippery slope toward open-source systems. Since open-source systems cannot be controlled, politicians would ultimately need to ban devices that are not controlled by the state. The end point: telescreens like those in Orwell’s 1984, devices that both monitor you and broadcast only the information approved by the state. One can only imagine the methods. Will the police stop and search people on the street looking for unauthorized phones? Prison sentences for buying a non-state computer on the black market? Charges of organized crime for smuggling in containers of open-source software on USB sticks? Welcome to a brave new world.

The Zero-Knowledge Proof alternative and the EU

While the rest of the world is moving forward with identity verification plans, the EU has presented its own privacy-focused approach to age verification. In April 2026, Ursula von der Leyen, President of the European Commission, unveiled an age verification app with “the highest privacy standards in the world” and the presentation materials describe the app as “completely anonymous.”

The EU app is open source, and the EU countries are supposed to use it to create their own versions, and become the issuers of the age credentials that their citizens can then use. This means we may see many different versions of age verification from member states (if they even choose the EU app at all; several countries have already said they prefer to develop their own independent solutions).

At its core, the EU app works like this: you, as the user, provide your identity to an issuer using something like an ID card. The EU envisions member states acting as issuers. The issuer then provides you with a number of credentials that you can use on websites and social media platforms to prove that you are old enough. These credentials only confirm that you meet the age requirement; they do not reveal your identity to the website or platform. Each credential is also used only once. For example, Facebook and X would receive different credentials that they cannot link together, meaning they cannot be used to build a pattern or profile of your internet behavior (and thereby identify who you are). This setup is why – we assume – the EU calls their app completely anonymous. There’s only one little problem. The issuer knows which credentials belong to which person. If you were to post something the state considers inappropriate on some platform, the state could ask the platform for the age credential and easily identify who it belongs to. The consequence is that you cannot post anonymously.

The solution to this problem is so called Zero-Knowledge Proof (ZKP) cryptography. With fully developed ZKP technology, you still need to provide your identity to the issuer, but the issuer would not be able to connect the credentials used on websites and services back to you. Right now, the EU app does not have ZKP functionality, contrasting Ursula von der Leyen’s claim that the app ”is technically ready to be used”. But more importantly, the app is currently designed to always function without ZKP technology; if ZKP is unavailable, the app falls back to a non-ZKP model. Even if fully developed ZKP technology could be implemented in the future, it would remain an optional extra feature that countries may choose to disable and that the EU could remove at any time.

This means that the EU could decide at any time that ZKP may no longer be used, and in one stroke the app would fall back to its default mode, meaning that every post on social media carries an ID tag. By that point, an infrastructure will already have been rolled out; people will have gotten used to it, and it will be harder to roll it back.

Age verification based on Zero-Knowledge Proof technology would be better than full on identity verification. However, even with fully functioning ZKP technology, age verification would still have significant problems. One issue is that people without ID documents would be excluded (determining age through facial recognition, for example, is not sufficiently precise). It would also enable states to revoke “problematic” individuals’ ability to express themselves online by refusing to issue age credentials. Another concerns a more fundamental question: is it really reasonable that young people should be entirely prevented from using social media? Is it desirable that 15-year-olds are completely blocked from expressing themselves on platforms that reach the public? Not all social media platforms are based on collecting data about everyone, manipulating users, and algorithmically steering them in various directions. An illustrative example is that the UK’s Online Safety Act may limit access to Wikipedia. The assumption that age verification would gradually expand and exclude young people from meaningful digital meeting spaces is not far-fetched.

Identity verification for social media – just another “what about the children” excuse to introduce mass surveillance and censorship.

Children’s “safety” has long been used by intelligence agencies and other authorities as a battering ram for introducing mass surveillance, especially in recent years. In the United States, attempts were made under the Kids Online Safety Act (KOSA) to introduce identity verification using children as the justification. The same applies to the UK’s Online Safety Act, where politicians repeatedly test whether they can get scanning of end-to-end encrypted communication approved. In the EU, authorities have tried (hand in hand with American tech companies and intelligence agencies) to introduce total mass surveillance through the scanning of all communication. During 2026, the EU will make another attempt through its Going Dark/ProtectEU project, where the goal is client-side scanning – in other words, government spyware on all devices.

The age verification rush must be slowed down, and politicians who do not want a society in which citizens live under total surveillance need to recognize the consequences of different types of legislation.

Aquileo | 2026 security assessment of our Android app

2026-06-01T10:26:11.381449+00:00

Our Android app has for the second time passed MASA, a standardized security assessment, conducted by Leviathan Security Group.

Following last year’s assessment we’ve recently conducted the Mobile Application Security Assessment (MASA) to further ensure our compliance with modern secure mobile app development. It checked version 2026.2 of our app against the Mobile App Profile (MAP) specification and identified a few minor issues. These issues were addressed in version 2026.3-beta3 (later released as 2026.3), which resulted in a pass for our app.

Overview of findings

The initial testing round identified six issues, of which one false-positive and one not applicable. Here’s an overview of the addressed issues that were also re-tested against version 2026.3-beta3.

1.5.1.4 All Pending Intents shall be immutable or otherwise justified for mutability

A few PendingIntents were incorrectly marked as mutable, however we do not believe it posed much risk to our users since the app has very limited intent capabilities.

Conclusion: We agree with the finding and the intents have been changed to immutable.

1.5.3.1 The app shall by default mask data in the User Interface when it is known to be sensitive

On the login screen the account number input was not hidden, and instead was shown in plain text. When adding or editing a custom API access method the password was also shown in plain text.

Conclusion: We agree that those inputs should be masked to protect against shoulder surfing attacks so we’ve updated the UI to hide the sensitive input by default.

1.8.2.1 The app shall be transparent about data collection and usage

After adding support for in-app purchases via Google Play, our data collection policy on Google Play was inadvertently overlooked. To enable refunds we store a link between a purchase and an account for 20 days, as described in our privacy policy, this applies to Play Store purchases as well.

Conclusion: Our Google Play listing has been updated with Purchase history in the Data collection section to be as transparent as possible.

1.8.3.1 Users shall have the ability to request their data to be deleted via an in-app mechanism

Our app did not provide an in-app mechanism to delete accounts. This was by design due to the way our app and service works. We don’t believe it adds much value but rather opens up for abuse or mistakes. Instead we have mechanisms to continuously delete the little data we have, e.g. the link between accounts and payments that’s needed to enable refunds. More about this in our privacy policy.

Conclusion: We’ve implemented in-app account deletion to meet the MAP specification.

Read the report

You can check out the official App Defense Alliance Directory entry here and see that the app is independently reviewed in the Google Play Store. Unfortunately Google has not published the certificate yet, but once available it will be directly accessible using this link. You can also check out a more technical summary as well as test reports and the compliance report in our GitHub repository.

Last words

We would like to thank Leviathan for the thorough assessment. The communication was professional, and the assessment was carried out to a high standard and provided us with valuable insights.

Aquileo | Exit IP fingerprinting between VPN servers

2026-05-20T15:12:05.203366+00:00

On Friday the 15th of May, we became aware of a fingerprinting issue affecting Mullvad users.

When a user switches from one VPN server to another, this sometimes makes it possible for services such as websites to confidently guess that the same user that connected from the new VPN server is the one that connected from the previous VPN server.

This does not reveal the identity of the user. It can however reveal the fact that someone that previously connected from one VPN server has now connected from another VPN server.

Fingerprinting is telling devices apart by looking at properties that make them unique or close to it. Fingerprinting is a problem in many domains. The Mullvad Browser and DAITA are examples of protections against fingerprinting in web browsers and traffic analysis.

How it works

Each VPN server has many users. For both IPv4 and IPv6 every user will be assigned one exit IP address on the server from which the user's traffic will be sent out to the internet. There are technical limitations to how many users can use the same exit address, which is why servers have a range of several exit addresses. Each user device has a unique WireGuard key used to encrypt the connection. There is also an internal tunnel address that is usually but not always correlated with the user’s WireGuard key.

The issue arises when connecting to different VPN servers with the same internal tunnel address. Then the user is likely to be assigned an exit address with the same relative position in each VPN server's range of exit addresses. If for example this is 40%, then it will be an exit address about 40% into in the range on all VPN servers.

Server A

1.1.1.1
1.1.1.2
1.1.1.3
1.1.1.4 <--
1.1.1.5
1.1.1.6
1.1.1.7
1.1.1.8
1.1.1.9

Server B

2.2.2.101
2.2.2.102
2.2.2.103
2.2.2.104 <--
2.2.2.105
2.2.2.106
2.2.2.107
2.2.2.108
2.2.2.109

Usually, lots of users are assigned to every exit address so this will not provide certainty but in many cases good guesses can be made.

What should I do?

Depending on your threat model, you only need to change your behavior if you change VPN servers specifically to stop the ability to link what you do on one server to what you do on another. In this case, our recommendation would be to log out and log in again in the Mullvad app if switching servers. This will regenerate the WireGuard key and change the internal IP address.

What is being done

Going forward, our new method to assign which exit IP addresses someone is using on one VPN server, will give no information on which exit address is used on another VPN server, or by another user on the same server. This change is currently being tested and is planned to start being rolled out to our VPN servers in the coming weeks. Progress updates will be available here.

Aquileo | Any app on recent Android versions can leak certain traffic

2026-05-12T09:16:16.798119+00:00

A recently discovered bug in Android 16 allows any app to leak traffic outside the VPN tunnel.

As reported in the post The Tiny UDP Cannon: An Android VPN Bypass, Android 16 introduced a bug that allows a malicious app to send traffic outside the VPN tunnel, including with “Always-On VPN” + “Block connections without VPN” turned on. This affects all VPN apps, not just Mullvad VPN. This blog post aims to spread knowledge about this issue to help keep our users safe.

Having traffic leak outside the tunnel means your real IP address becomes visible on the Internet, which could potentially be used for tracking or surveillance purposes.

The bug was reported to the Android Security Team, but was closed as Won’t Fix (Infeasible), as described in the linked article. After consulting with the report author (https://x.com/cybaqkebm), we reported the issue on the Android issue tracker. However, at the time of writing the issue is marked as inaccessible by Google for unknown reasons.

In contrast, GrapheneOS, a security-focused Android-based OS, quickly patched the issue in its codebase.

Technical details

The leak involves calling the registerQuicConnectionClosePayload method on the ConnectivityManager system service. The purpose of this method is to enable graceful teardowns of QUIC connections by sending a final payload to the server, letting it know that the device has closed the connection. However, this function does not properly check that the payload is sent into the VPN tunnel, resulting in an attacker being able to craft a payload that is sent outside the VPN, thus leaking the device's real IP address.

Mitigation

A mitigation is possible, but is quite technical in that it requires USB debugging to be enabled on the device in order to run the following Android Debug Bridge (adb) commands:

adb shell device_config put tethering close_quic_connection -1
adb reboot

This disables the QUIC graceful shutdown feature, and thus closes the leak. The mitigation will persist across reboots, but it may be undone by system updates, in which case the steps will need to be repeated.

Performing this mitigation means that the server-side QUIC socket will remain half-open until it times out, which should generally not negatively affect the Android device or apps running on it. However, only use the command at your own risk if you understand the implications.

Conclusion

This is not the first and certainly not the last leak of its kind in Android. You can protect against it by applying the mitigation above or by running a security-oriented Android variant such as GrapheneOS. Ultimately, though, it also comes down to not running apps you do not trust on your devices.

Aquileo | Force all app traffic into the tunnel

2026-04-21T14:03:52.167021+00:00

A year ago, we wrote about how bugs in Apple's networking stack are preventing the iOS app from being as secure as possible. The bugs are still there, but we have secured our app anyway.

The Problem

Due to the intricacies of Apple's NetworkExtension framework, we have been stuck with a VPN app that we knew would leak traffic in some circumstances on iOS. There is a known fix for this but it comes with significant downsides. The biggest one being the way this breaks the app update and the user get stuck in a broken update loop.

AppStore determines it should update our app
iOS bricks the networking stack whilst trying to update the app
User reboots phone
Phone can reach the internet again
AppStore determines it should update our app
iOS bricks the networking stack whilst trying to update the app

The workaround

We have decided that we are not going to wait anymore and we would like to offer our users the best possible privacy and security, even if it comes with major UX limitations.

With more users experiencing these limitations first hand we also increase the likelihood that the issue will be resolved upstream.

Thus, soon we will be releasing a new version of the iOS app that will contain a feature called Force all apps. Under the hood, enabling this feature sets the includeAllNetworks configuration option to true. We have tried to make sure that users who enable the feature do so deliberately, without making them jump through too many hoops. The phone can still enter the broken update loop, but now users should receive a notification about a new version being available before the app gets auto-updated.

Updating the app

You must use one of these methods to avoid getting stuck in an update loop.

Disconnect the VPN while the app is updated.
App will not reconnect automatically after it is updated, but Force all apps will remain enabled.
Disable Force all apps while the app is updated.
App will reconnect automatically after it is updated, but Force all apps will have to be re-enabled manually.

In both cases, your traffic will leak during the update process - we do not believe there is a workaround for this.

We do however expect a minority of our users using this feature will end up with a broken networking stack, and unfortunately there is not much we can do. If you've been affected by this, we can only encourage you to capture the anguish and express it as a feedback report to Apple.

As for the bug where, with includeAllNetworks enabled, our tunnel process cannot bind sockets to the tunnel device - that is still there and it still is an issue. As such, our workaround to use userspace networking will remain in place.

Aquileo | Mullvad Browser Alpha moves to Firefox Rapid Release and adds Linux ARM support

2026-03-26T12:43:51.423969+00:00

Starting with 16.0a1 alpha release, Mullvad Browser Alpha is based on the Firefox Rapid Release channel rather than the Extended Support Release (ESR). The alpha release is now available on Linux ARM.

Every year, to correspond with a new Firefox ESR release, there would be many months of stressful work in order to catch up with a full years worth of Firefox changes.

Tor Project is now experimenting with a different approach to releases, spreading the workload of the annual ESR transition throughout the year. To this end, Mullvad Browser Alpha has transitioned to the Firefox Rapid Release channel, while the Stable channel remains on Firefox ESR.

What this means for Mullvad Browser Stable users

We have previously had two feature releases per year: one in Q2 and one in late Q3. With this new development model, we expect to have only one major feature release per year.

What this means for Mullvad Browser Alpha users

This shift brings several changes that users of the Alpha channel should be aware of:

Quicker Access to Features: Users will receive new upstream features from Mozilla shortly after they are introduced, rather than waiting for the next major ESR update.
Potentially Less Secure and Private: Because features will be shipped more rapidly, the likelihood of encountering upstream bugs that could impact security and privacy increases. As always, Alpha builds are intended for testing purposes. If you are an at-risk user or require a reliably working browser, you should use the Stable channel.
A Less Predictable Release Cadence: Sometimes the intersection of upstream changes, our build system, and our patches introduce rather difficult problems for us to solve. It may take longer than the available four week window of time between scheduled Rapid Release versions to implement fixes.

For a comprehensive explanation of the technical rationale, you can read the full announcement on the Tor Project’s blog: The Future of Tor Browser Alpha.

Availability on Linux ARM

We are also happy to announce Mullvad Browser Alpha is now available for ARM platforms on Linux. You can download it manually via our website or install it directly using our RPM and DEB repositories.

Become a tester!

Now is a great time to become a Mullvad Browser Alpha tester! However, if you are at risk or need strong anonymity, please stick with Mullvad Browser Stable.

Aquileo | A security audit of GotaTun is now available

2026-03-06T09:50:01.594564+00:00

In December we wrote about our new WireGuard implementation GotaTun, which is currently used in our Android app. An independent audit by Assured Security Consultants is now available.

The audit was performed between 19th January and 15th February. It encompassed all of GotaTun v0.2.0 with the exception of DAITA, the GotaTun CLI, and external dependencies. No major vulnerabilities were found. Two issues with a low severity rating were identified, along with several notes. Their findings were summarized as follows:

Based on our code review, GotaTun has no major vulnerabilities. However, the parts where it deviates from the WireGuard specification (as described in Observations 3.2 and 3.3) are cause for concern. Furthermore, the code contains some instances of ”TODO” comments which should be addressed.

The report included some recommendations, most of which were implemented before the audit was published including fixes for both low-severity issues and some of the informational notes. For the remaining notes we decided that they did not require immediate attention.

You can find the full audit here. You can also find our more in-depth summary and response to the audit here.

Main findings

3.2 LFSR used to generate peer identifiers (LOW)

In v0.2.0, 24 bits of the WireGuard session identifier associated with a peer were static for that peer, the remaining 8 bits being a predictable counter which increased with every new session (typically once every other minute). This deviated from the WireGuard specification which recommends generating a random 32-bit integer for every session.

The behavior was inherited from BoringTun. As was pointed out, it likely did not provide much information to a passive observer except potentially the number of handshakes that a peer has made or the number of peers active at any time. This could potentially be inferred from other information such as handshakes, IP addresses, and ports already. Nonetheless, we patched the code to conform to the WireGuard specification.

3.3 Padding of payload not according to WireGuard specification (LOW)

According to the WireGuard specification, packets should be padded before encryption so that their lengths are divisible by 16. The GotaTun source code did include a “TODO” comment about this, which is still present in BoringTun. We updated the code to always pad the payload before encrypting it for correctness, and because the extra padding may complicate traffic analysis slightly.

3.4 Endpoint address only updated on handshake initiation (NOTE)

If the IP address of a peer changed, GotaTun did not correctly begin sending packets to the new address in most cases. In other words, GotaTun did not handle roaming of peers correctly.

This did not affect the Mullvad VPN app per se, as our VPN servers never change IP addresses during an active WireGuard session. We still decided that the issue was important enough to fix in order to make GotaTun usable as a general purpose WireGuard implementation.

Closing words

We would like to thank Assured for this audit. We are grateful for their thorough comparison with the WireGuard specification and available reference implementations. All aforementioned fixes are generally available as of GotaTun v0.4.0, and our apps will automatically begin using it in coming releases.

GotaTun was released in our Android app last year, with promising results. With the audit complete, we have even more confidence in the reliability of GotaTun and plan to roll it out across all remaining platforms during 2026.

Aquileo | Addition of new supported currencies

2026-02-03T10:30:59.711475+00:00

As a way to improve the options our users have for payments and transaction fees, we have added new currencies when paying with a credit card!

It is now possible to pay using more local currencies such as United Arab Emirates Dirham, Mexican Peso and Turkish Lira when making a card payment.

These are in addition to offering payments with the nine currencies available for cash payments, the four cryptocurrency methods (which all offer a 10% discount when used) and a selection of other payment methods.

We continue to offer (and have no plans to change) our fair pricing model of €5 per month regardless of how long you pay for. If you are not satisfied we have a 14-day money back guarantee for several of our payment options, excluding cash and crypto.

Our Pricing Page has information about the current currency conversion rates for USD, GBP, SEK, AUD and CAD, as well as information about alternate places to buy our scratch cards from.

For the universal right

Mullvad VPN.

Aquileo | New security audit of account and payment services

2026-01-21T13:19:53.140177+00:00

Late last year X41 D‑Sec GmbH performed a white‑box source‑code audit of the Mullvad payment and account API and its supporting backend services.

The engagement covered the parts of the system that handle authentication, device provisioning, payment processing and the distribution of WireGuard keys.

Read the full audit report here.

What the audit found

The auditors identified five security‑relevant findings – two low‑severity and three medium‑severity issues – together with a number of informational notes. Importantly, none of the reported issues give an attacker the ability to access user data, nor do they weaken the privacy guarantees that Mullvad promises to its users.

The most notable issue is a voucher race condition that could let a single voucher be applied to multiple accounts. This affects billing only and does not expose any personal information.

Two of the medium findings were redacted to avoid publishing details that could cause availability issues. Those redactions do not hide any vulnerability that would compromise user privacy.

The remaining “informational” notes flag a variety of hardening opportunities – from improving mTLS usage across internal services to simplifying the Nginx configuration and signing the relay list – all of which we are looking into to further strengthen the overall security posture.

Redactions – why some details stay private

Four of the findings were redacted from the public version of the report. Those items describe potential ways of causing availability issues. They do not affect the confidentiality or integrity of customer data.

Looking back – the previous audit

This audit builds on the work presented in our earlier Security audit of account and payment services (2023). The current 2025 audit confirms that our account and payment service continues to hold up under close scrutiny, while also highlighting areas where modern best practices can be applied.

Aquileo | Mullvad Review of 2025

2025-12-30T11:57:24.923967+00:00

2025 has been a year of campaigns, audits and improvements to our suite of obfuscation methods.

Here we highlight what we think you should be aware of from this year as it comes to an end.

Quantum-resistant tunnels on every desktop platform

The first app release of the year, 2025.2 enabled Post-Quantum WireGuard by default on Windows, making it the final platform to enable quantum resistance by default. All apps across all platforms now have WireGuard with quantum-resistance enabled by default.

Partnership with Obscura VPN

February brought the announcement that we were commencing a partnership with the “Two-Party VPN” provider, Obscura VPN. This partnership means that customers of Obscura (a service separate from Mullvad VPN, and not available to Mullvad VPN customers directly) are able to use Mullvad’s WireGuard VPN servers as an “exit hop”. Further information is available on Obscura VPN's website.

Fighting for privacy with our first outdoor campaign of 2025

In late February we covered billboards, public transport and outdoor spaces of various cities across the USA with our stances on privacy, the debate on freedom of speech, and mass surveillance. This was the first of many campaigns we did during 2025.

Android security audit, reproducible builds and multihop

Our Android app was successfully audited during March, passing Mobile Application Security Assessment (MASA) in the process, as confirmed by NCC Group. We were proud to say that version 2024.9 of our Android app passed all controls without the need for any fixes or modifications.

March also brought the introduction of multihop to Android, version 2025.1 introduced the ability to route traffic through two servers.

Finally, the 2025.2 release of the Android app introduced reproducible builds, which means that anyone can verify that the app downloaded and installed is built from the open-source code which we publish.

Lightning payments

At the start of August we enabled instant transactions, lower fees and enhanced privacy with the introduction of Bitcoin Lightning. We run our own lightning node, and continue to offer a 10% discount when paying with it, along with all other cryptocurrencies supported by Mullvad.

Website security audit by Assured AB

The audit showed that we had no critical, high, or medium-severity issues on our website. All low-severity issues were promptly fixed during the audit period, and Assured concluded that “Good security practice is followed in all parts of the reviewed web applications”.

QUIC and LWO obfuscation methods

Two new methods to make it harder for firewalls to detect and block traffic were introduced in September and November respectively. Both obfuscations are available on a subset of our WireGuard servers.

Version 2025.8 on iOS and Android brought QUIC.

Android release 2025.9 and Desktop release 2025.13 brought LWO. iOS is planned for early 2026.

The introduction of these obfuscation methods brings the total to four added methods in 2025, which we mentioned as being part of our ongoing effort to fight censorship and improve the user experience in restrictive networks.

Leta shutdown

We decided to shut our search engine Leta down at the start of November. Leta was introduced as an experiment, a way to see whether we could improve the privacy of our users further. We concluded after two years that similar privacy can be achieved through the combination of a VPN and a privacy-focused browser.

Removal of OpenVPN from our desktop apps

OpenVPN will be completely removed on 15th January 2026, and in preparation, version 2025.14 of the desktop app removed all mention of the protocol. Please ensure you are prepared for the removal of OpenVPN prior to the cut-off date.

And then?

Chat Control is back on the menu, and to highlight the corruption behind the proposal we introduced our video campaign, “And then?”. This campaign has been split into multiple videos of varying lengths, shared across many different channels, online and on TV in Australia, Sweden, Germany, and the USA.

Shortly after the release of the campaign, we learned that the Council of Ministers in the EU had, after three years, reached a common position on Chat Control. The requirement for mandatory scanning (including end-to-end encrypted messaging services) was removed, which is a major victory.

And then?

We will continue to fight against Chat Control, our views on privacy, striving to make mass surveillance as impractical as possible throughout 2026 and beyond.

Mullvad VPN

Aquileo | Announcing GotaTun, the future of WireGuard at Mullvad VPN

2025-12-19T10:26:18.147618+00:00

GotaTun is a WireGuard® implementation written in Rust aimed at being fast, efficient and reliable.

GotaTun is a fork of the BoringTun project from Cloudflare. This is not a new protocol or connection method, just WireGuard® written in Rust. The name GotaTun is a combination of the original project, BoringTun, and Götatunneln, a physical tunnel located in Gothenburg. We have integrated privacy enhancing features like DAITA & Multihop, added first-class support for Android and used Rust to achieve great performance by using safe multi-threading and zero-copy memory strategies.

Last month we rolled it out to all our Android users, and we aim to ship it to the remaining platforms next year.

Why GotaTun?

Our mobile apps have relied on wireguard-go for several years, a cross-platform userspace implementation of WireGuard® in Go. wireguard-go has been the de-facto userspace implementation of WireGuard® to this date, and many VPN providers besides Mullvad use it. Since mid-2024 we have been maintaining a fork of
wireguard-go to support features like DAITA & Multihop. While wireguard-go has served its purpose for many years it has not been without its challenges.

For Android apps distributed via the Google Play Store, Google collects crash reports and makes them available to developers. In the developer console we have seen that more than 85% of all crashes reported have stemmed from the wireguard-go. We have managed to solve some of the obscure issues over the years (#6727 and #7728 to name two examples), but many still remain. For these reasons we chose Android as the first platform to release GotaTun on, allowing us to see the impact right away.

Another challenge we have faced is interoperating Rust and Go. Currently, most of the service components of the Mullvad VPN app are written in Rust with the exception of wireguard-go. Crossing the boundary between Rust and Go is done using a foreign function interface (FFI), which is inherently unsafe and complex. Since Go is a managed language with its own separate runtime, how it executes is opaque to the Rust code. If wireguard-go were to hang or crash, recovering stacktraces is not always possible which makes debugging the code cumbersome. Limited visibility insight into crashes stemming from Go has made troubleshooting and long-term maintenance tedious.

Outcome

The impact has been immediate. So far not a single crash has stemmed from GotaTun, meaning that all our old crashes from wireguard-go are now gone. Since rolling out GotaTun on Android with version 2025.10 in the end of November we’ve seen a big drop in the metric user-perceived crash rate, from 0.40% to 0.01%, when comparing to previous releases. The feedback from users' have also been positive, with reports of better speeds and lower battery usage.

User-perceived crash rate

Looking ahead

We’ve reached the first major milestone with the release of GotaTun on Android, but we have a lot more exciting things in store for 2026.

A third-party security audit will take place early next year.
We will replace wireguard-go with GotaTun across all platforms, including desktop and iOS.
More effort will be put into improving performance.

We hope you are as excited as we are for 2026!

Aquileo | Final reminder for OpenVPN removal

2025-12-15T13:40:38.216472+00:00

This is the final reminder that we are fully removing support for OpenVPN on January 15th 2026, in a month's time.

This means we will no longer have any OpenVPN servers in a month.

We blogged about this in November 2024, also here when we removed OpenVPN support from our desktop apps.

If you are still using OpenVPN in any way, we strongly advise that you switch to WireGuard.

We have guides on how to use WireGuard in the help section of our website.

OpenVPN servers will continue to work until 15th January 2026, but new servers will not be added, and existing servers will be taken offline as the days go by.

It will not be possible to generate new OpenVPN configurations soon.

WireGuard is the Future

For the universal right to privacy.

Aquileo | Removing OpenVPN from the Mullvad VPN app

2025-12-10T10:00:46.713446+00:00

In version 2025.14 of the desktop app, we removed the ability to select OpenVPN as a tunnel protocol.

We announced the decision to retire OpenVPN last year, continuing our adoption of WireGuard that started in 2017. OpenVPN has provided a valuable anti-censorship alternative for many users, but with more WireGuard anti-censorship methods available in the app WireGuard now provides the same level of obfuscation with superior security and performance.

How to migrate to WireGuard

The transition to WireGuard will happen automatically and will not require any further action. If you had previously selected OpenVPN as the VPN protocol to use, it will automatically transition to WireGuard.

Below are some potential issues that might occur if you had OpenVPN selected:

Location

If you had previously selected a specific OpenVPN server, or a location / custom list containing only OpenVPN servers, your internet connection may be blocked until a new location is selected. You may need to choose a different location as OpenVPN servers are no longer available.

Note that OpenVPN servers will no longer show up in custom lists.

Anti-censorship

In some cases, OpenVPN has been able to bypass firewalls and censorship where WireGuard could not. To restore similar functionality, go to:

“Settings” -> “VPN Settings” -> “Anti-censorship”

To mimic OpenVPN over TCP, select “UDP-over-TCP” and set port to 443.

To mimic OpenVPN with Bridge mode, select “Shadowsocks”.

Selecting either of these anti-censorship methods will wrap the WireGuard traffic in an additional layer of obfuscation that resembles the OpenVPN protocol.

The default option, “Automatic”, will try to select an appropriate method on your behalf when plain WireGuard fails to connect. Manually selecting another option can reduce the time it takes to connect by using the specified option immediately.

Be aware that enabling any anti-censorship method does not improve your privacy. If you are in a situation where you can use our service without obfuscating your traffic, there is no need to change any obfuscation settings. Enabling “UDP-over-TCP”, “ShadowSocks” or "QUIC" can have a detrimental impact to the performance and resource consumption of your device.

MSSfix

If you previously used the OpenVPN “Mssfix” option to limit packet size, consider adjusting the WireGuard “MTU” setting accordingly. Go to “Settings” -> “VPN Settings” and scroll to the bottom to find it.

External VPN apps and routers

If you have configured an external app or router to connect to Mullvad OpenVPN servers, then it will stop working on or before 15th January 2026 when the last servers will be taken down. The option to generate new OpenVPN configurations from your account page may cease to exist before then.

Aquileo | An important victory – but we still need to stop Chat Control.

2025-11-26T17:32:14.233811+00:00

An important victory - but we still need to stop Chat Control.

The Council of Ministers in the EU has, after three years, now reached a common position on Chat Control. The requirement for mandatory scanning (including end-to-end encrypted messaging services) has been removed, which is a major victory. The EU Council failed to implement mandatory mass surveillance. However, in its proposal, they are laying the groundwork for mass surveillance in the future.

What happens now?

The Council will now enter negotiations with the European Parliament, led by the European Commission. We urge the Parliament to stand firm in the trilogue negotiations and not deviate an inch from its previous position, demanding: no mass surveillance whatsoever without suspicion and a court order, no ID-verification requirements, and no censorship of legal content.

The EU Council is preparing for mandatory mass surveillance and censorship

The Council’s version of Chat Control includes voluntary scanning, vaguely worded legislation that may entail requirements for age verification and mandatory ID checks (even for end-to-end encrypted services), and an article stating that the requirement for mandatory scanning shall be reconsidered every three years. They also introduce a new infrastructure for blocking material, where it is up to each member state to block what they consider illegal. At the same time, a massive EU center is being established to work exclusively on this. All in all, this indicates that the EU Council is aiming to build an infrastructure for mass surveillance, and the legislative proposal is written in a way that opens the door to it.

The EU Council’s Chat Control version

The EU Council’s Chat Control version introduces a new type of scanning for so-called new material and grooming. This means that AI will scan people’s conversations, photos and videos, in search of criminal content. This will result in enormous numbers of false positives, and people’s private lives will move from an AI detection to being examined by employees at a new EU center. This is mass surveillance and people’s private lives will be scanned without any suspicion and without a court order. This scanning is carried out in cooperation with American companies and can at any time be used to scan for virtually anything; Europol has
already requested broader scanning and wants access to material that is not illegal.

Every three years, the European Commission will challenge the law and attempt to force mandatory scanning (even for end-to-end-encrypted services).
Messaging services (including end-to-end encrypted) must take “all reasonable measures” to reduce the risk of their services being misused, including implementation of age verification. This means that the EU may require ID checks and ban anonymous use of messaging services and social media. This poses problems for people who criticize those in power in authoritarian countries, for whistleblowers who want to leak documents, and for sources who wish to speak anonymously with journalists.
A new infrastructure for blocking material is introduced, where it’s up to each of the member states to issue blocking orders for what they consider illegal. This implies that content that is illegal in one country will also be blocked in a country where it is legal. Once this infrastructure is in place, it also opens the door to a slippery slope when it comes to censorship.

Stop Chat Control

From the outset, Chat Control was a proposal that aimed to introduce mass surveillance. That ambition is clearly still present within the Commission and among many of the member states in the Council. The Council failed to introduce mass surveillance but has succeeded in paving the way for new attempts. This applies not only to future proposals for mandatory chat control scanning every three years. This is part of a broader development in which private and secure communication is being challenged by forces seeking to introduce mass surveillance. ProtectEU is a rebranded Chat Control, aimed at banning encryption. National laws are trying to do the same. We need to put a stop to these attempts here and now.

Aquileo | Mullvad VPN presents And Then?

2025-11-14T10:46:37.220560+00:00

Chat Control is back on the menu. To highlight the corruption behind the proposal, Mullvad VPN now presents "And Then?"

Update: On November 26, 2025 the EU Council, after three years, agreed on a common position on Chat Control.

Chat Control is once again back on the menu. In the Council of the EU (the member states), several countries continue to work on new versions of the bill. The latest draft in November 2025 was presented with different branding and different semantics, but it would result in mass surveillance, AI-scanning of private data, ID requirement to use messaging services and – with vague legislative text – risk of mandatory scanning (even for end-to-end-encrypted services) in the future.

As long as the Council refuses to reject the bill (the way the European Parliament did), the Chat Control proposal could still become law – despite violating EU law and fundamental human rights.

To highlight the effects of mass surveillance and remind people of the corrupt history (full story below) behind the Chat Control proposal, we now present the film "And Then?"

Your browser does not support the video tag.

And then: short versions.

These are broadcast in EU countries, to oppose the Chat Control proposal. They are also broadcast in other countries, where mass surveillance has gone too far, to remind people of the absurdity of systems where people are monitored without warrants and without suspicion of a crime.

Your browser does not support the video tag.

It’s time to end the Chat Control proposal and other dishonest initiatives like ProtectEU. We don’t need more mass surveillance. On the contrary, we need to move away from dragnet systems and bulk data collection. This goes for countries all over the world, if they want to be free and open societies.

Demand transparency from your politicians, and privacy for the people.

The backstory

First, Ashton Kutcher (yes, the actor) convinced the EU Commission that they could scan everything on an EU citizen’s phone or computer (messages, photos, emails, phone calls, all of it) for child sexual abuse material without, at the same time, looking at the content of other types of communication. This could be done in a secure way, Kutcher told European politicians, and he would know (sure), since he was running a tech organization called Thorn.

And then?

And then EU Commissioner Ylva Johansson presented the legislative proposal called Chat Control, which aims to scan everything on every EU citizen’s phones and computers (including conversation in end-to-end-encrypted messaging services). The message from the Commission was: we will only search for child sexual abuse material (CSAM).

And then?

And then the proposal was slammed by both the Council of Minister’s and the European Commission’s own legal service, as well as by the European Parliament’s Data Protection Board. The UN Human Rights Council described Chat Control as incompatible with fundamental human rights and stated that the proposal would lead to mass surveillance and self-censorship. Former judges at the European Court of Justice said that the proposal was in breach of the EU Charter of Rights and 465 researchers joined forces to warn of the consequences.

And then?

And then Ylva Johansson claimed that anyone calling the proposal “mass surveillance” was just trying to smear it. Experts from all over the world explained to her that the kind of scanning she was talking about (as Ylva described it: a drug-sniffing dog that can detect illegal content in a message without reading the message) simply cannot be done safely, and that Chat Control would mean the end of privacy and pose a security threat to all Europeans. Ylva responded with: “what about the children?”

And then?

And then it was revealed that Thorn, the organization founded by Ashton Kutcher and which had been lobbying for Chat Control from the beginning, was selling the kind of scanning technology that could be used for Chat Control – despite being registered as a charity organization in the EU’s lobbying registry.

And then?

And then it was revealed that Thorn, together with the EU Commission, had also started and funded “children’s rights organizations” that had publicly supported the proposal. On the boards of these children’s rights organizations sat the EU Commission, non-European intelligence services, and representatives from Thorn. What appeared publicly to be charitable organizations were in fact lobby groups that – according to internal documents – operated in line with this strategy: “once the EU Survivors Taskforce is established and we are clear on the mobilised survivors, we will establish a list pairing responsible survivors with Members of the European Parliament – we will ‘divide and conquer’ the MEPs by deploying in priority survivors from MEPs’ countries of origin.”

And then?

And then it was revealed that Europol wanted unlimited access and wanted to use the scanning for more than just child abuse crimes, saying that all data – also unfiltered and innocent material – should be stored because it “could at some point be useful to law enforcement”.

And then?

And then politicians in Brussels wanted to exempt themselves from the scanning.

And then?

And then the European Parliament, in an almost historic consensus, voted against the proposal and called Chat Control nothing but mass surveillance. As one of the members of the parliament said: “The Commission wasn’t focusing on protecting children but wanted mass surveillance.”

And then?

And then Ylva Johansson’s office tried to win the battle in the Council of the EU (law proposals must go through both the Parliament and the Council) by using illegal micro-targeting on social media. Ylva Johansson was summoned to a hearing in the European Parliament and questioned about Thorn's involvement in the law proposal, the illegal micro-targeting, and the Commission's unwillingness to release public documents regarding the relationship between Thorn and the EU commission. Ylvas answer: think of the children.

And then?

And then the EU Commission realized that Chat Control was slipping out of their hands. So, they launched the initiatives Going Dark and ProtectEU – projects once again aiming to gain lawful access to people’s private conversations and data. They had abandoned the excuse “think about the children” and now went broader, claiming the surveillance is crucial to stop organized crime.

And then?

And then the fight continues for private and secure communication in the EU (and other parts of the world as digital communication is global). The EU Council is still trying to agree on a common position on Chat Control (more than three years have passed and they still haven’t reached one). If Chat Control won’t become law, we can expect ProtectEU to be the next attempt. Once again, non-European intelligence services are involved in the process, and the arguments are the same ones used by Ashton Kutcher and Ylva Johansson: if we don’t get access to everything, we are going dark (this expression comes originally from the FBI).

And then?

And then it’s up to the people of Europe to make their voices heard. Otherwise, we risk ending up with a system that means total mass surveillance, massive security threats and zero privacy for the vast majority of people in the EU (as well as other parts of the world).

And then?

Aquileo | Introducing Lightweight WireGuard Obfuscation

2025-11-12T09:38:10.770722+00:00

We are now adding Lightweight WireGuard Obfuscation (LWO), aimed at helping users bypass firewalls and censorship. This new obfuscation method is now available on all desktop platforms and Android.

LWO (Lightweight WireGuard obfuscation) makes it harder for firewalls to detect and block VPN traffic. With this update, our app should become more usable in countries and networks where WireGuard traffic and our other obfuscation methods are otherwise blocked.

How to Enable LWO Obfuscation

To use the new LWO obfuscation, make sure you have at least version 2025.13 of the desktop Mullvad VPN app or 2025.9 on Android. The iOS app will include LWO obfuscation in a future release.

With the default settings, the app will automatically try LWO after a few failed connection attempts. You can configure the app to always use LWO obfuscation by following the instructions below.

Go to Settings → VPN Settings → WireGuard Settings → Obfuscation → LWO.

Or run the following terminal command on desktop: mullvad obfuscation set mode lwo

LWO is a fast and lightweight obfuscation protocol that builds on top of WireGuard. It works by scrambling the header of each WireGuard packet, making them harder to fingerprint as VPN traffic. This is a very computationally cheap operation, which means that LWO adds very little overhead on top of WireGuard and is thus fast to perform even on low-powered devices. This results in great throughput performance and lower power consumption compared to other obfuscation methods such as Shadowsocks.

This update brings our fourth WireGuard obfuscation and the second one this year after launching QUIC obfuscation in September. It is part of our ongoing effort to fight censorship and improve the user experience in restrictive networks.

Aquileo | Shutting down our search proxy Leta

2025-11-06T12:36:43.362496+00:00

On November 27, 2025, we will shut down our search proxy, Leta.

Leta's primary benefit was that it acted as a privacy proxy for search. Pooling and caching requests on behalf of a great number of users.

The search industry continues to undergo big changes. Leta will not be able to follow and will likely become less useful over time.

Similar privacy can be achieved through the combination of a VPN and a privacy-focused browser.

We have therefore decided to discontinue Leta and continue to advance the development of state-of-the-art of VPNs and browser privacy - through our own work and research and in collaboration with our partners.

Aquileo | Independent security audit of our web app completed by Assured

2025-10-23T08:12:03.541238+00:00

Assured completed an independent security assessment of our web app. The assessment found no critical, high, or medium-severity issues. One low-severity input validation weakness was identified and promptly fixed.

A recent third‑party assessment found no issues enabling crashes or unauthorized data access. We remediated one input‑validation weakness and addressed five non‑security observations; one framework‑related behavior was accepted with no security impact.

The application previously lacked length limits on certain fields, allowing unusually large inputs (~500k+ characters). While this did not allow crashes or unauthorized access, it could cause error messages to echo raw input and consume unnecessary resources. We implemented strict application‑layer validation and sanitized error handling to mitigate both outcomes.

Security reviews are integral to our privacy commitments: strong security underpins all our privacy-by-design services. Read the full report here. Thank you to Assured for the thorough assessment. We will continue to conduct regular independent audits and publish transparent results.

Aquileo | QUIC Obfuscation now available on Android and iOS

2025-10-20T03:19:06.845311+00:00

We are pleased to announce that our latest obfuscation feature, QUIC, aimed at helping users bypass firewalls and censorship, is now available on Android and iOS.

Our QUIC obfuscation tunnels WireGuard traffic, making it harder for firewalls to detect and block it. With this update, our app should become more usable in countries and networks where WireGuard traffic and our other obfuscation methods are restricted or blocked.

How to Enable QUIC Obfuscation

To use the new QUIC obfuscation, make sure you have at least version 2025.8 on Android or 2025.8 on iOS.

With the default settings, the app will automatically try QUIC after a few failed connection attempts. You can configure the app to always use QUIC obfuscation by following the instructions below.

Go to Settings → VPN Settings → Wireguard obfuscation → QUIC.

You can read more about QUIC and our QUIC obfuscation here.

This feature is only required if you have trouble connecting, or are located in a restrictive location. There are no performance or privacy benefits of enabling this feature.

Aquileo | Introducing QUIC Obfuscation for WireGuard

2025-09-09T11:26:00.513860+00:00

We are excited to add QUIC obfuscation for WireGuard, aimed at helping users bypass firewalls and censorship. This new obfuscation method is now available on all desktop platforms.

How to Enable QUIC Obfuscation

To use the new QUIC obfuscation, make sure you have at least version 2025.9 of the desktop Mullvad VPN app. The Android & iOS apps will include QUIC obfuscation in future releases.

With the default settings, the app will automatically try QUIC after a few failed connection attempts. You can configure the app to always use QUIC obfuscation by following the instructions below.

Go to Settings → VPN Settings → WireGuard Settings → Obfuscation → QUIC.
Or run the following terminal command:mullvad obfuscation set mode quic

QUIC is a fast and lightweight transport protocol based on UDP. It is aimed at replacing the traditional use of TCP in HTTP/2, and is being adopted rapidly as the web transitions to HTTP/3.

What we call QUIC obfuscation builds on the MASQUE protocol described by RFC 9298 - Proxying UDP in HTTP. As the title of the RFC implies, QUIC obfuscation works by tunneling UDP through an HTTP server acting as a proxy. For a censor looking at the traffic being sent between a client and server, the traffic will appear as web traffic. HTTP is generally not blocked by state-level censors, since much of the internet would be unreachable without it.

This update brings yet more resilience on top of the fast & secure WireGuard VPN protocol. We hope this feature enhances your experience, especially in restrictive networks. Give it a try, and see if it works for you, we would love to hear your feedback!