Security researchers at Mysk published findings in 2026 about a subtle iOS privacy weakness: apps on your iPhone can read the private IP address assigned inside your VPN tunnel, and use it as a fingerprint to track your VPN session. It is a niche issue, but a real one, and it exposed a genuine difference between how VPN providers build their iOS apps.
What “internal tunnel IP fingerprinting” means
When you connect to a VPN, your device gets two addresses. There is the public IP the outside world sees, which is the server’s, and there is a private IP assigned inside the tunnel to your device. That internal address is not supposed to be interesting. It is plumbing.
The problem Mysk identified is that on iOS, apps can read that internal tunnel IP, which they cannot do the same way on other platforms. If your provider hands you a recurring private IP that stays the same or predictably rotates, an app can use it as a marker to recognise your VPN session across time, even as your public IP changes. That is the fingerprint.
To be clear about scale: this is not your browsing being exposed to the internet, and it is not your public identity leaking. It is a tracking vector that apps already installed on your phone could exploit to correlate sessions. Serious for high-risk users, minor for most people, but worth understanding.
Where providers differ
Mysk’s tests found that Proton VPN sidesteps the problem by design. Proton assigns every user the same internal tunnel IP, so there is nothing unique to fingerprint. If everyone looks identical inside the tunnel, an app reading that address learns nothing that distinguishes you.
An earlier version of the reporting claimed Proton was the only provider to prevent this. That was too strong, and it got corrected. The research focused on WireGuard implementations and did not account for custom protocols. ExpressVPN clarified that its default Lightway protocol uses a similar fixed internal IP allocation, which mitigates the same tracking vector, while its WireGuard implementation rotates IPs on reconnection. So the honest picture is: some providers are protected by design, at least on their default protocol, and others are more exposed depending on how they assign internal addresses.
Want to compare all VPNs side by side? Check our full VPN comparison table with scores across 18 criteria.
How worried should you be?
For most people, this is low on the threat list. The attacker here is an app already on your phone that wants to correlate your VPN usage over time. That is a real capability, but it is not the same as your traffic being decrypted or your public IP leaking to a website. If your threat model is “I do not want advertisers and ordinary apps building a profile,” it is relevant. If it is “I want Netflix from another region,” it is not.
For high-risk users, journalists, activists, people under surveillance, it matters more, because session correlation is exactly the kind of metadata that can deanonymise someone over time. Our guide for journalists and high-risk users goes into this class of threat.
The practical takeaway is that architecture choices you never see, like how a provider assigns an internal IP, can have real privacy consequences. It is another reason the provider you trust should be one that thinks about these edges. In our comparison, Proton VPN scores 4.24/5 with a perfect 5/5 on leak protection, and this research is a good example of why that score reflects more than marketing.
What you can do
You cannot patch iOS yourself, but you can pick a provider whose default protocol does not hand out a unique, trackable internal IP. Proton VPN is the clearest example from the Mysk research, since it assigns everyone the same internal address. If you already use another provider, check whether its default protocol uses a fixed internal allocation, and prefer that over an implementation that rotates unique IPs on iOS.
Beyond this specific issue, the usual hygiene still applies: keep iOS updated, run a genuine no-logs provider, and test for leaks. Our guide to checking for VPN IP leaks walks through the standard checks, and our best VPN for privacy ranking covers the providers that take architecture seriously.
The Mysk research is a useful reminder that iOS leaks more metadata than users assume, and that internal tunnel IPs can fingerprint a VPN session. It is a low priority for casual users and a real concern for high-risk ones. Proton VPN handles it by giving everyone the same internal IP, and ExpressVPN's default Lightway mitigates it too, so it is not a "one provider" story. Pick a provider whose default protocol does not hand you a unique, trackable internal address, keep iOS updated, and test for leaks.