> The Government says that "systemic" weaknesses cannot be demanded.
That's because they already have a pre-existing "systemic" weakness that's better than any encryption back door: automatic software updates. If you can replace the software so it gives you the unencrypted data why on earth bother with a breaking the encryption? All they need is a hammer that forces the software companies to write undetectable bugs and silently install them for them, and that's what this legislation provides.
We handed this systemic weakness to them on a platter, and it's been there for years now. Even through it was many years ago, I still recall the horror I felt when my daughter has her phone stolen, and I discovered I could press a button on Google play that would install some spy software to report on the whereabouts of the thief. I'm sort of surprised it took them this long to wake up to it.
It isn't impossible to fix, so I suspect in the long term this loophole will be closed. The key to the fix is in the word "systemic", which translated means someone other than them can't exploit the weakness.
As an aside, they are apparently operating under the assumption they will be able to control who has access to it. Which is to say they believe they can control access to something that will be highly automated thus ultimately controlled by only a few people. They are after all subject to the same attack they are using on us - they will be asking programmers to update their software, software that they undoubtedly will never see because it's "company proprietary". As the saying goes every human has their price. The price the attackers can afford in this case is extraordinary: this system is the key that unlocks every banking password, every bank SWIFT password, every GPG key, every X509 secret key, every email, every boardroom discussion on billion dollar takeovers. They are kidding themselves if they think that can protect this - which is why it is a terrible proposal.
Worse, they don't have the defence we do, which is that the "normal" unmonitored population must be running strong, secure software. We get this unbugged software now from public servers we call app stores. So all you need is something that will compute the hash of the software you downloaded so you can compare it to publicly available one, and won't lie about the result. "Won't lie about the result" translates to "a device that can't be corrupted" which in turn translates to "can't have it's software upgraded". We already have such devices: they are called TPM's. We already know how to use them. Sort of. They work real well in 2FA dongles for example.
Nevertheless it has to be said the primary application of TPM's, secure boot, hasn't been a raging success. But then we haven't had a good reason to make it a success: how many people do you know have been victims of evil maid's? Well, that was nice while it lasted, but now we all about to come face to face with an evil maid from our worst nightmares: someone who can install software updates while your phone is sitting in the safety of your coat pocket without leaving a trace.
So the incentive is now here, the engineering task is well defined. Unfortunately the problem remains hard. We have to surround drivers, IPC, network stack with same high Chinese walls we currently put around apps, and somehow tie this all back to an all seeing TPM. So it's going to take a while. Maybe se4L will get it's day in the sun.
That's because they already have a pre-existing "systemic" weakness that's better than any encryption back door: automatic software updates. If you can replace the software so it gives you the unencrypted data why on earth bother with a breaking the encryption? All they need is a hammer that forces the software companies to write undetectable bugs and silently install them for them, and that's what this legislation provides.
We handed this systemic weakness to them on a platter, and it's been there for years now. Even through it was many years ago, I still recall the horror I felt when my daughter has her phone stolen, and I discovered I could press a button on Google play that would install some spy software to report on the whereabouts of the thief. I'm sort of surprised it took them this long to wake up to it.
It isn't impossible to fix, so I suspect in the long term this loophole will be closed. The key to the fix is in the word "systemic", which translated means someone other than them can't exploit the weakness.
As an aside, they are apparently operating under the assumption they will be able to control who has access to it. Which is to say they believe they can control access to something that will be highly automated thus ultimately controlled by only a few people. They are after all subject to the same attack they are using on us - they will be asking programmers to update their software, software that they undoubtedly will never see because it's "company proprietary". As the saying goes every human has their price. The price the attackers can afford in this case is extraordinary: this system is the key that unlocks every banking password, every bank SWIFT password, every GPG key, every X509 secret key, every email, every boardroom discussion on billion dollar takeovers. They are kidding themselves if they think that can protect this - which is why it is a terrible proposal.
Worse, they don't have the defence we do, which is that the "normal" unmonitored population must be running strong, secure software. We get this unbugged software now from public servers we call app stores. So all you need is something that will compute the hash of the software you downloaded so you can compare it to publicly available one, and won't lie about the result. "Won't lie about the result" translates to "a device that can't be corrupted" which in turn translates to "can't have it's software upgraded". We already have such devices: they are called TPM's. We already know how to use them. Sort of. They work real well in 2FA dongles for example.
Nevertheless it has to be said the primary application of TPM's, secure boot, hasn't been a raging success. But then we haven't had a good reason to make it a success: how many people do you know have been victims of evil maid's? Well, that was nice while it lasted, but now we all about to come face to face with an evil maid from our worst nightmares: someone who can install software updates while your phone is sitting in the safety of your coat pocket without leaving a trace.
So the incentive is now here, the engineering task is well defined. Unfortunately the problem remains hard. We have to surround drivers, IPC, network stack with same high Chinese walls we currently put around apps, and somehow tie this all back to an all seeing TPM. So it's going to take a while. Maybe se4L will get it's day in the sun.