How big threat do you think Intel ME is in reality, not in theory?
When it comes to Intel Management Engine, I actually think it's not a threat if you neutralize it. I mean to just set the HAP bit on it. Because if that isn't enough then that means all computers in the world which use Intel CPU can be accessed by NSA but if NSA had this much power then it seems obvious that they aren't using it and why wouldn't they use it?
There's a github project to neutralize/disbale Intel ME: https://github.com/corna/me_cleaner
Disable is overwriting intel ME as much as possible with zeros, leaving only a little remaining to be able to boot the computer.
The newer the intel chips are, the less likely it is to be able to disable it.
But all chip sets can be neutralized which means to set the HAP bit which is an official feature. In theory we can't actually trust the HAP bit to really disable intel ME permanently. It's more like asking Intel to do what they have promised because it's proprietary. But I think it really does permanently disable it because otherwise NSA would be abusing this power.
That's why I think the newer laptop models are better because it's probably not necessary to disable, it's enough to just neutralize withthe HAP bit. And with a newer modern laptop they can have open source Embedded Controller firmware which is better than proprietary Embedded Controller firmware.
I choose not to think about it or include it in my mental threat model, the same way I choose to not worry about thermonuclear warheads.
If there’s some exploitable backdoor and Intel gets owned, we’re all boned and there’s nothing we can really do about it. I don’t have anti-ballistic-missile systems, and I also don’t have the capability to make an entire hardware/firmware/os from scratch.
So instead focus on the things you can control and are more likely to happen. Don’t plan for doomsday, plan for every day.
I would go on eBay and buy a Libreboot machine from 2009 and prior (X200, T500, etc.) These systems have 100% no blobs in the firmware and can have the IME fully disabled. I use these as my daily and I'm fine.
IMHO Intel ME or the AMD equivalent are only relevant for state level targeted attacks.
It wouldn't be wise for them to waste it on the small fries and risk having some snoopy I-have-nothing-better-to-do-with-my-life security researcher find some attack payloads.
Of course you are right to be worried and think about it. Right now the best you can do is coreboot, it allows you to disable it.
If you want to counter that risk the best is to get a computer like the nitropads (coreboot and only open source firmware, qubeos on top) https://www.nitrokey.com/news/2020/nitropad-secure-laptop-unique-tamper-detection or the ones of system76
After that, it's no use worrying too much. You could as well be hit be hit in a car crash, a seism or a tsunami could also hit you city. Don't think about it too much, just have a small plan so you are not too lost if the black swan comes for you.
Open source is not enough. It needs to be entirely free software. I recommend buying a Libreboot laptop from before 2009, they can fully disable/remove the IME and have a 100% free BIOS firmware (anything supported device with a Core Duo processor basically).
Thanks!
I dug in and just found out that you can buy libreboot computers with Intel ME disabled and support the libreboot project on https://minifree.org/
Buying other hardware that you (well... not me ;) can inspect and verify, e.g RISC?
For now the performances are pretty terrible BUT one can imagine, assuming they have the right discipline and mental model doing what's actually personal on a verifiable processor, e.g browsing and reading emails, and what's not, e.g watching a TV show on another machine with CPU/GPU with an unverifiable architecture.
PS: I have a Precursor and a Banana Pi BPI-F3 with SpacemiT K1 8 core RISC-V chip and that's the main idea behind them both, i.e knowing, as a community, how it works all the way down.
The NSA tries incredibly hard to not make public which of the many many options in their toolbox are in active use at any given time. Not sure anyone outside the org can say for sure what they are and aren't using.
Just restrict network access (both in and out) with proper (trusted) hardware firewall. It's much safer that relying on disabling / configuring etc. You can't attack what you can't reach (directly or in reverse).
I don't have experience with that yet. Are you talking about a PI hole?
Can you give a little idea on how to make such firewall rules? Because I want to have a laptop with many VMs or Qubes and each VM has different firewall rules. An email qube would only allow connection to the email server. Maybe one of the safe browsing VMs would only allow connections to the websites I typically visit. The unsafe VM maybe to everything except for known bad IPs/domains.
And NSA and other potential adversaries most likely have access to at least one domain that isn't blocked by firewall.
Crowdstrike didn't target anyone either. Yet, a mistake in code that privileged, resulted in massive outages. Intel ME runs at even higher privileges, in even more devices.
I am opposed to stuff like kernel level code, exactly for that reason. Mistakes can be just as harmful as malice, but both are parts of human nature. The software we design should protect us from ourselves, not expose us to more risk.
There is no such thing as a back door that "good guys" can access, but the bad guys cannot. Intel ME is exactly that, a permanent back door into basically every system. A hack of ME would take down basically all cyber infrastructure.
I genuinely believe the NSA et al is targeting those who attempt to avoid all targeting more than the average person. It's difficult to avoid being tracked, it's nearly impossible to additionally blend in with an unsuspicious façade. Might as well become a secret agent if you're capable of avoiding the NSA's gaze.