What if every computer is infected with an undetectable virus?

>attacker has somehow gotten a script with root permissions running on my device whenever it's on >attacker modifies the PGP binary so that it allows their fake signatures of the code to get through >attacker can now implant their own versions of the kernel, browsers, etc. by pushing fake kernel updates >attacker can also now fake any webpage they want by rewriting the local CA tree, thus allowing them to implant .iso files of every operating system, so that any new installation media I make will also be infected >attacker may even write a worm that spreads this way, sitting in infected devices for months or years >victim can never even check with other methods that source code or binaries were modified from the original because attacker can modify the sha256sum and openssl binaries and the piece de resistance: >imagine someone exploited a 0day on the machine of a major distro's kernel package signer and got root, meaning that they can freely modify a software update, and they can also arrange for the original to be lost in a catastrophic disk failure so it's nearly impossible to detect the worm at all >this attack could have happened any time in the past decade or two on, say, debian. Who could remember? >it is very likely that every debian machine in existence has been infected with this virus >it is very likely that every distro's kernel package maintainer would at some point used a debian machine to sign an update >tfw every linux machine is probably infected with an undetectable virus that could be controlled by a single entity. The Windows one would fundamentally work the same way except the Windows people probably don't have one singular package signer person, you'd just have to figure out *some* way to get a hack running on machines right when they go out -- possibly by rooting a lazy (or "patriotic"...) computer manufacturer's systems.

This kind of Domesday scenario just isn't realistic. You're talking about *multiple* *major* nation-state powers collaborating. It's easier to use court orders, since everything of value is online and owned by morally dubious opportunists.

>>18427 The malware described in the first half would take a massive amount of work to develop for very little gain. The attacker already has root and they're going to go through all this trouble just for persistence? They either want to steal your info which can be done quickly enough without persistence or they want your machine for a botnet/mass crypto mining setup. Your individual machine on its own is worth very little to the entire botnet not all that trouble. If they were attacking like a company or something and wanted to persist for a long time it would be a lot less work (and more effective) to just move laterally through the network to a machine that's not getting its OS reinstalled. Also maintainers compare package binaries to binaries compiled from source for QA and security reasons before sending it off to the repos. It's not that breaking into repos and filling them with malicious packages isn't a thing, people have tried it before. But as they found out it's just really hard to do because repos are inherently community run so you have to fool everyone in the community or else it doesn't work.

>>18427 >What is Intel Management Engine? >What is AMD Platform Security Processor?

>>18427 Someone post info about poor security at computer factories where they put viruses on new computers before they get packed... Also fedex said my new laptop was held up in chinese customs for 3 days <,<

>>18427 If you are paranoid to the point where you are seriously treating this as anything other than a fun hypothetical, then I have no idea why you would still use technology at this point. Like >>18432 said, its such a high effort low gain scheme where one small mistake would blow the entire plan wide open. However, I believe it to be worth mentioning that closest thing to compromised repos was this https://www.bleepingcomputer.com/news/security/malware-found-in-arch-linux-aur-package-repository/ tl;dr all it really did was log system data. Not sure if anything came from this afterwards (and if someone has info on it please post it here) but do with this info what you will.

>>18433 >>18427 The OP's doomsday scenario is *hypothetically* possible in the form of Intel ME or AMD PSP, which are not auditable and have not been fully reverse engineered or otherwise analyzed yet. So far, there are no known backdoors (besides previously patched nightmare scenario security exploits lol). A backdoor is hypothetically possible, but consider the difficulty of hiding a true keys-to-the-kingdom backdoor - there would need to be multiple ways of triggering the rootkit, probably with different implementations for each operating system. It would be very difficult to build and ship a backdoor of this level of access clandestinely. The agencies who do this sort of thing also understand the cost of a backdoor falling into the hands of someone else, which will inevitably eventually happen. As a result, I find the idea that these processor supervisors contain rootkits to be mostly implausible. The truth is that all operating systems (yes, ALL of them, Linux, BSD, Winblows, etc) have fundamental design flaws that ensure there will always be vulnerabilities, so there isn't even a need to ship with pre-installed backdoors in the first place! If the above bothers you, you can buy a Purism laptop with coreboot, install Qubes, and compartmentalize everything you do. I am sure that there are some Xen zero-days stockpiled by state adversaries, but that's about as well as you can do, sorry!

This is just the Ken Thompson hack https://wiki.c2.com/?TheKenThompsonHack

>>18427 This type of Doomerism just isn’t productive. Even if you build you’re own TempleOS you would still have to worry about hardware backdoors like intel ME. You have to trust someone else’s code eventually, question is do you trust the people who let you read the code you’re running or the people who don’t?

>>18427 Very possible. But as others have mentioned, your scenario is only likely for a group of sufficient resources. What's more possible is the also-mentioned assembler/compiler backdoor. For example, an assembler backdoor might prepend a malicious stub between a label and the instruction it references: or even prepend the malicious stub before the main body of code. This would be a better target overall, because it's more general than independently cracking the many operating systems in use today. Even more damaging is the also-mentioned hardware backdoor. A completely-homebrewed machine should solve this. Builder beware. >>18440 >This type of Doomerism just isn’t productive. I disagree. This type of Doomerism can be a positive step forward. The attack/defense game is inherently asymmetric, and makes "perfectly securing our systems" an exercise in failure. Even so, this doesn't mean that the set of all "perfectly secure" systems is empty. There remains an unknown probability that we will eventually discover each of them.

As long as there are no backdoors on the hardware side, you can bootstrap your software from MS-DOS to a reasonably modern Linux distribution in a matter of days. That's called bootstrapping, anon.

>>18429 Yeah, thats the Occam's Razor of mini botnet everywhere. Theres enough info about anyone to get a picture, and all glowies collaborate. A finer picture isnt even necessary for crimes against nation states. One would have to start leaving less trace 5 years before they start the crime. The only organization that I know of that has more information than governments, including all of the above-ground governments information, are the 40+ species of aliens that operate here. The aliens already know I got cheese pizza, but the above-ground government doesnt know. Yet. Alexander

>>18461 >The aliens already know I got cheese pizza, but the above-ground government doesnt know. The Discordian Cabals know, we just can't decide what to do with the information

>>18461 so much for improving the quality of the board lol

>>18433 They're (((spyware))), you should already know this

>>18462 You people can't do anything with that intel because youre understaffed and drunk all the time. Alexander

>>18469 >can't handle one post that isnt against him, somewhere out in the fringes of the internet Thats why you aint great.

Reminds me of this https://www.teamten.com/lawrence/writings/coding-machines/

>>18429 >domesday That was the day I turned 23, accepted my hair loss, and started shaving my head

>>18598 that's funny, i started taking testosterone blocker and estrogen instead.

>>18602 word so if you start losing your hair it'll be a desperate situation. male pattern baldness doesn't look girly. but the estrogen will take care of that won't it? won't it? lol

>>18461 >One would have to start leaving less trace 5 years before they start the crime. Not so. Drop any semblance of long-term pseudonymity, since that is the bullet which strikes the 99% of cybercriminals, and choose to indulge either the bandwidth overhead or higher latency of the so-called "Anonymity Trilemma", combined with a bit of wireless or radio indirection for your first hop for membership concealment, and no global passive adversary has a chance of _retroactively_ deanonymizing an attacker.

it's possible. It's called proprietary software.