Very uninformed person here and a genuine question. Isn’t TPM endorsed by respected security projects such as GrapheneOS, I mean the Titan chip isn’t some type of TPM equivalent for computers and one of the main reasons GrapheneOS doesn’t support other phones that aren’t Pixel?
Trusted computing and TPMs aren’t inherently bad. Like all issues of trust, it comes down to who the trusted parties are and what they’re asking of you.
So for example, let’s start with the idea of a work computer. Say you work for a bank and they issue you a laptop. In order to access all the sensitive data related to a bank, certain guarantees must be made about the environment. The hard drive must have full disk encryption (FDE) so that if it’s ever lost or stolen, the information that may have been on it can’t be compromised. This is not your laptop. This is not your environment. This is for the most part, totally fair.
Now let’s consider Microsoft and your personal device. Microsoft is forcing you into their trusted environment by requiring online accounts and TPM/SecureBoot. And how do you benefit? FDE through BitLocker, sure. But you know there are other FDE solutions and BitLocker results in you losing control of your keys because they are automatically uploaded through your online account to Microsoft for “recovery” purposes. ~Source~Related What Microsoft is really saying here is that they have a trusted environment, and if you are to be a trusted party in that environment with the “privilege” of accessing their software and services, you must submit your personal device to their rules. Are you starting to feel the icky vibes here?
This is made worse by TPM 2.0 supporting remote attestation.
This of course raises the question, verified to what degree and to whose standards? Are they simply trying to protect us from maliciously crafted software, or is it DRM to prevent running pirated content, Trusted Platform Module (TPM) 2.0 and Secure Boot for Call of Duty. Of course this is ostensibly for anti-cheat, but you see how quickly that moves adjacent to other purposes. How much are you willing to give up to maintain (a sense of) security?
EDIT: One final point. Trusted systems are the general security engineering concept of protecting systems through enforced policies to achieve certain levels of trust. Trusted Computing is a very specific set of technologies with a board of directors worth taking a quick look at …
You can use the TPM to automatically decrypt a LUKS root volume at boot just like you would BitLocker, however your recovery keys aren’t automatically uploaded to a Microsoft account, you must manage them yourself (generally I see this as a benefit but the layman may appreciate Microsoft’s “assistance” here). https://wiki.archlinux.org/title/Trusted_Platform_Module
⚠️ WARNING, what follows is much more my personal speculation on things so absolutely take this with a grain of salt.
The TPM isn’t ever really under the user’s direct control - it’s used by applications that hook into it. On Linux, I anticipate you would be much more protected from the remote attestation aspects of TPM 2.0 phoning out to 3rd party servers for verification because in general that just does not vibe with the FOSS standards and sensibilities. HOWEVER, in my wildest speculations it may still be possible to fall victim to that through proprietary software. Currently things like Microsoft Office, Adobe Photoshop, or Activision’s Call of Duty don’t work under Linux. If Microsoft gets particularly desperate, I wouldn’t put it past them to actually distribute a native Office for Linux package, or work with Adobe or Activision to do likewise for their programs as a baited hook. Any proprietary, closed-source software can still communicate with the exposed TPM for that remote attestation and refuse to run if they find tampered data, pirated files, or other running applications they object to (I don’t know exactly what form it would take but it could be any or all of these). Effectively they maintain control over your system by right of denial; if you want to run their software you play by their rules.
This of course doesn’t matter if you have no desire to run that software. Again, the TPM itself is not directly malicious and as long as you don’t engage with software that would use it maliciously, it’s fine to have it active and enabled within your OS.
The rant is because I’m trying to provide a balanced view of it without coming off as a fearmonger. TPM is certainly not without its uses, but it’s a leash that can be yanked on. Under Windows, you’re fully in Microsoft’s world and they will yank that leash. But given the right leverage and circumstances, that leash can and very well may extend into Linux as well if you allow the software through with it.
Be careful. Use it if you will but remember what it is capable of.
The thing is, trusted computing as a security feature isn’t useless. For the particular case of phones, people generally use relatively low entropy passwords, because it’s impractical to do otherwise. The Titan chip uses trusted computing technologies to ensure that an attacker with physical access cannot bruteforce the password, which it does by forcing a timeout between successive attempts. It might do other things too, this isn’t my area of expertise, but (I believe) it isn’t needed for the general functioning of the device, as opposed to e.g. the intel ME.
Of course, a security chip that you have the power to control would be better, and no less secure, but that doesn’t exist. However, neither the OS itself not the apps directly depend on trusted computing. Otherwise GrapheneOS couldn’t exist in the first place.
Note that this problem doesn’t exist on desktops or laptops: it is entirely possible to memorize a passphrase around 96 bits of entropy, which is high enough that it can’t practically be bruteforced, especially if the algorithm to check if it is correct is computationally slow.
So, you lost a bit of sovereignty for your phone in the interest of security, but phones aren’t private to begin with: the actual modem also uses trusted computing. The devs behind GrapheneOS considered this the best solution to the problem, after weighing the pros and cons. Personally, I’d be happy to have a flip phone which has no password, and then do everything of significance (possibly including call and SMS) on my laptop. That is to say, I’d rather I didn’t have to use GrapheneOS, but it’s compromises align well with my own for now.
Very uninformed person here and a genuine question. Isn’t TPM endorsed by respected security projects such as GrapheneOS, I mean the Titan chip isn’t some type of TPM equivalent for computers and one of the main reasons GrapheneOS doesn’t support other phones that aren’t Pixel?
Trusted computing and TPMs aren’t inherently bad. Like all issues of trust, it comes down to who the trusted parties are and what they’re asking of you.
So for example, let’s start with the idea of a work computer. Say you work for a bank and they issue you a laptop. In order to access all the sensitive data related to a bank, certain guarantees must be made about the environment. The hard drive must have full disk encryption (FDE) so that if it’s ever lost or stolen, the information that may have been on it can’t be compromised. This is not your laptop. This is not your environment. This is for the most part, totally fair.
Now let’s consider Microsoft and your personal device. Microsoft is forcing you into their trusted environment by requiring online accounts and TPM/SecureBoot. And how do you benefit? FDE through BitLocker, sure. But you know there are other FDE solutions and BitLocker results in you losing control of your keys because they are automatically uploaded through your online account to Microsoft for “recovery” purposes. ~Source ~Related What Microsoft is really saying here is that they have a trusted environment, and if you are to be a trusted party in that environment with the “privilege” of accessing their software and services, you must submit your personal device to their rules. Are you starting to feel the icky vibes here?
This is made worse by TPM 2.0 supporting remote attestation.
This of course raises the question, verified to what degree and to whose standards? Are they simply trying to protect us from maliciously crafted software, or is it DRM to prevent running pirated content, Trusted Platform Module (TPM) 2.0 and Secure Boot for Call of Duty. Of course this is ostensibly for anti-cheat, but you see how quickly that moves adjacent to other purposes. How much are you willing to give up to maintain (a sense of) security?
EDIT: One final point. Trusted systems are the general security engineering concept of protecting systems through enforced policies to achieve certain levels of trust. Trusted Computing is a very specific set of technologies with a board of directors worth taking a quick look at …
Are we saying here that using Linux + TPM = recommended, using Microsoft + TPM = burn?
Mostly, kind of.
You can use the TPM to automatically decrypt a LUKS root volume at boot just like you would BitLocker, however your recovery keys aren’t automatically uploaded to a Microsoft account, you must manage them yourself (generally I see this as a benefit but the layman may appreciate Microsoft’s “assistance” here). https://wiki.archlinux.org/title/Trusted_Platform_Module
You can also use it for SSH, https://www.ledger.com/blog/ssh-with-tpm
⚠️ WARNING, what follows is much more my personal speculation on things so absolutely take this with a grain of salt.
The TPM isn’t ever really under the user’s direct control - it’s used by applications that hook into it. On Linux, I anticipate you would be much more protected from the remote attestation aspects of TPM 2.0 phoning out to 3rd party servers for verification because in general that just does not vibe with the FOSS standards and sensibilities. HOWEVER, in my wildest speculations it may still be possible to fall victim to that through proprietary software. Currently things like Microsoft Office, Adobe Photoshop, or Activision’s Call of Duty don’t work under Linux. If Microsoft gets particularly desperate, I wouldn’t put it past them to actually distribute a native Office for Linux package, or work with Adobe or Activision to do likewise for their programs as a baited hook. Any proprietary, closed-source software can still communicate with the exposed TPM for that remote attestation and refuse to run if they find tampered data, pirated files, or other running applications they object to (I don’t know exactly what form it would take but it could be any or all of these). Effectively they maintain control over your system by right of denial; if you want to run their software you play by their rules.
This of course doesn’t matter if you have no desire to run that software. Again, the TPM itself is not directly malicious and as long as you don’t engage with software that would use it maliciously, it’s fine to have it active and enabled within your OS.
So, what is the point of the TPM “rant” if it has great use?
Well, I wouldn’t say great, merely useful.
The rant is because I’m trying to provide a balanced view of it without coming off as a fearmonger. TPM is certainly not without its uses, but it’s a leash that can be yanked on. Under Windows, you’re fully in Microsoft’s world and they will yank that leash. But given the right leverage and circumstances, that leash can and very well may extend into Linux as well if you allow the software through with it.
Be careful. Use it if you will but remember what it is capable of.
The thing is, trusted computing as a security feature isn’t useless. For the particular case of phones, people generally use relatively low entropy passwords, because it’s impractical to do otherwise. The Titan chip uses trusted computing technologies to ensure that an attacker with physical access cannot bruteforce the password, which it does by forcing a timeout between successive attempts. It might do other things too, this isn’t my area of expertise, but (I believe) it isn’t needed for the general functioning of the device, as opposed to e.g. the intel ME.
Of course, a security chip that you have the power to control would be better, and no less secure, but that doesn’t exist. However, neither the OS itself not the apps directly depend on trusted computing. Otherwise GrapheneOS couldn’t exist in the first place.
Note that this problem doesn’t exist on desktops or laptops: it is entirely possible to memorize a passphrase around 96 bits of entropy, which is high enough that it can’t practically be bruteforced, especially if the algorithm to check if it is correct is computationally slow.
So, you lost a bit of sovereignty for your phone in the interest of security, but phones aren’t private to begin with: the actual modem also uses trusted computing. The devs behind GrapheneOS considered this the best solution to the problem, after weighing the pros and cons. Personally, I’d be happy to have a flip phone which has no password, and then do everything of significance (possibly including call and SMS) on my laptop. That is to say, I’d rather I didn’t have to use GrapheneOS, but it’s compromises align well with my own for now.