I haven’t heard of this being an issue with Mullvad or iVPN, but if someone who uses macOS could test them, that would be helpful.
Because the vast majority of users are going to use the official client. Even if you don’t agree ProtonVPN should be removed, there should at the very least be a disclaimer stating that the client kill switch has major problems on macos and linux.
Do you not…go outside?
Only with road warrior VPN setups (Tailscale, ZeroTier, etc) is this feasible.
Your disagreement is with PG’s existing minimum criteria for VPNs.[1] I agree that the criteria is too stringent to be a minimum qualifier, especially since the criteria also covers support for 3p clients (which may have a proper killswitch).
Minimum to Qualify: Kill switch built in to clients … Private VPN Service Recommendations and Comparison, No Sponsors or Ads - Privacy Guides ↩︎
Is this site recommending VPNs based on potential future software updates?
I would say that this kill switch issue is a significant betrayal of trust, especially since this is not a one-off bug, but rather a decision that they have made, which they admittedly have been lying about for a long long time. On their official website they claim that the kill switch does block your internet when switching between ProtonVPN servers - yet, when confronted on reddit, they deleted the post and eventually backtracked on the claim without updating the official website.
Considering the advanced discussions this forum has on protecting your IP address and the stringent requirements Privacy Guides has for recommending services, I personally cannot see how ProtonVPN on MacOs can be recommended.
Would you accept using a VPN that has this issue? I’m guessing you are not a ProtonVPN user on Mac - in any case, we all have different standards, I would like Privacy Guides to make this issue clear on their recommendations page.
Sure.
If only there was a line of routers specifically meant for travel…some sort of travel router hmmm
My impression is that they do not want to invest the time. I have not had this issue with MullvadVPN. In any case, if there were a technical limitation they should openly state this instead of lying on their website. As a user, I was shocked to discover this was happening.
That’s fine to think that, but then Privacy Guides should say so. Currently it says “If a VPN provides their own custom client, we require a kill switch to block network data leaks when disconnected.” - So anyone reading the recommendation would rightly think this is the case for ProtonVPN, alas it is not, and at a very minimum I do think this should be made clear.
I think it is ironic that ProtonVPN states on their wireguard config guide : “We strongly recommend using our official app on your Mac. It provides the easiest way to connect to our servers and allows you to benefit from many of Proton VPN’s advanced features. For example: Kill switch”
Not everyone has router access unfortunately… In any case, this router argument deviates from the point at hand.
It was really meant to just be an aside. Kind of regret even mentioning it now.
Not sure if they are outright lieing but proton does seem to consistently fudge the truth about their products such as calling Lumo open source.
Fair. As i said before, i agree that Proton does not technically meet the criteria. I would just prefer the criteria be changed then proton be removed.
This is a direct quote from their support page on kill switch:
“Please note that our regular kill switch feature can’t protect you if you intentionally disconnect from a VPN server. However, the feature does protect you while switching servers with Proton VPN.”
Then, you can read their response on reddit:
“To clarify, Kill Switch on Mac will not prevent your device from connecting to the internet during manual disconnection events [in context, they are referring to switching ProtonVPN servers] as you’ve described above. For this, you would need Advanced Kill switch, available on Windows and Linux only”
Their support page doesn’t fudge anything, it is pretty clear what they mean to say, and based both on user reporting and their admission in the reddit post, this is a lie.
I can get behind this if it’s not standard VPN behavior on macOS. If the kill switch is less trustworthy than the minimum requirements demand, and this is due to a flaw in Proton’s implementation rather than a platform limitation, it seems clear cut it should be removed.
This is probably a separate topic but on Linux if you really need a kill switch you can create a new network namespace with only the VPN interface attached and run your apps inside it as a workaround.
This topic is temporarily closed for at least 4 hours due to a large number of community flags.
This topic was automatically opened after 4 hours.
… “in the future” is not enough …
Ex:
Protons kill switches on Mac (and iOS) are useless IMO. They use the “always on” feature that is integrated into Apple’s VPN implementation. It will attempt to reconnect immediately if your WiFi drops, but will NOT block your connection if you change servers.
Mullvad and IVPN have much more robust and functional VPN applications on Mac with properly functioning kill switches.
I’m not voting to remove this because I think this is a fine product for probably a lot of people, especially people with a proton subscription already who do not have a super high threat model.
But 100% a warning needs to be added about where the kill switches are working and where they are not. That goes for all the VPNs.
Maybe PG should split the VPN recommendation category into two more specific silos
VPN Service Providers: reliable, privacy-centric server networks. Wireguard support, multi-hop, etc. Proton continues to excel here
VPN client software: the actual application users run to send traffic through a VPN tunnel. Reliable kill switch, open source, cross-platform (maybe) etc. The aforementioned issue with ProtonVPN’s kill switch falls under here
PG currently groups them together under a singular VPN recommendation page, presumably bc most trusted VPN service providers also provide a VPN client. But one could make the argument that these are two very different products, and should be evaluated independently
Just tested the kill switch on Linux with the wireguard framework in the Proton app. Kill switch seemed to be working just fine. Agreed that if you’re so concerned and your threat model is such that your free VPN isn’t good enough, use the Wireguard client.
Not all distros will be created equal, and, as shocking at it might seem, not all VPN users are torrenting everything in the universe.
Please read the thread. To summarise: ProtonVPN doesn’t meet minimum criteria laid out by PG, as its macOS client has a broken killswitch.
Not the Linux client.
As far as I know, for the official WireGuard client on Linux, the “killswitch” has to be set separately. It isn’t built-in.
Not removing it or not changing the minimum criteria for VPNs right away then makes PG a “dangerous” source of information. I mean, we won’t tolerate this from other projects or orgs, so why Proton?
To test the linux leak you have to follow the instructions here:
I use ProtonVPN on MacOS and iOS, using WireGuard CLI on Mac and the WireGuard app on iOS. Anything to worry about here?
WireGuard clients don’t have built in kill switches. I hope you knew that. But otherwise nothing wrong with using WireGuard as a VPN protocol for your obfuscation needs.
I agree. I think that if Privacy Guides had to absolutely make a change, this would be the solution.