Store Now Comprehend Later

I recently have been thinking about the store now decrypt later threat model, where an adversary stores your encoded messages until they can discover a vulnerability to decrypt them. With quantum computers starting to get large enough to break classical encryption, this has become a growing concern. However, there is still the classic anti-privacy argument that you have nothing to hide or that the feds wouldn't bother decrypting you because it isn't worth the human personnel. The problem is, now that large language models are starting to get better, you wouldn't need a dedicated human to pore over someone's search history, you could just have a neural net read someone's entire history on the web and sum them up on demand. I think this is probably the most plausible threat to privacy because we pretty much know that all of the elements are already here. There isn't any reason to believe it won't happen. >The federal government has been tapping most internet communication and storing it for years >The feds are ready and willing to conduct mass surveillance and breach privacy, and have the lack of accountability and funds to actually pull it off at scale. >The quantum computers we know about are close to having enough qubits to break encryption, and the government would have strong strategic incentive to keep a capable computer a secret. >For all of their flaws, neural nets are pretty good at categorizing large datasets by complex criteria. TLDR; The feds could take your tapped data, decrypt it with quantum computers, and then analyze it at scale using neural nets. Is it over privacybros?

store now decrypt later is a valid concern quantum computers is not (lightyears away) meme ai is not (emperor has no clothes) practise good persec/opsec, nothing will ever be perfectly secure forever but it's far from over

>>22546 It is silly to think that the feds don't already have a quantum computer capable of breaking encryption. In a decade or so, when information gets out that they had it all along, everyone will think that in hindsight it was obvious. My thesis is that it is either too expensive to operate, and thus impractical to crack everyone's communications, or they're worried that if they start using the quantum computer willy nilly, everyone will start to suspect a working QC, and so they're using it only on special occasions. On the bright side, there are already encryption algos that claim to be "Quantum-proof"

>>22547 >quantum computers are lightyears away The publicly known "cutting edge" is a few orders of magnitude too small by most estimates, but like >>22553 said if the alphabet boys had one they wouldn't tell anyone about it. If it wasn't a viable threat, why would the government be hosting quantum safe encryption competitions now. >meme AI isn't there I agree that most of the AGI and GPT soykaf is a massive grift, in fact I think that "artificial intelligence" is broadly a misnomer, which is why I don't mention it in the op. What I am talking about is purpose built natural language processors that go through your call and internet history, and prepare a summary profile of you on demand. Flags you for certain beliefs, etc. With a data set the size of all internet traffic and training examples from agent composed profiles, I see no reason why current neural net technologies wouldn't be up to the task. Give the glowies a couple years and everything you did that wasn't quantum safe will be neatly summarized and categorized. I hope you haven't done anything you wouldn't want on your permanent record in the past 10 years.

What most privacy dum dums don't understand is that there are quantum resistant encryption methods, and for understanding that you need higher math knowledge

>>22555 Tell me o wise one, are you securing your internet traffic with quantum resistant methods? Do you use the quantum safe version of TOR?

>>22553 >too expensive to operate Black projects of intelligence agencies are funded by tax payers. They have ALL the cash. >>22556 No, because I am not interested in security

>>22554 >If it wasn't a viable threat, why would the government be hosting quantum safe encryption competitions now. There are a few reasons 1) There could be a huge breakthrough N years from now that suddenly makes quantum computers practical 2) it's good to have public key algorithms with hardness not based on discrete log or prime factorization, which have consistently gotten weaker over time *classically*. 3) Cryptographers like to have *huge* security margins. AES-256 would take thousands of years to break with all the mass-energy of the earth converted into a perfectly efficient quantum computer. 4) The security theater industry can make a killing off it

>>22561 Easy publications too ? > first xxxxx from post quantum assumptions

>>22562 Yes that is sort of the academic version of point 4.

>With quantum computers starting to get large enough to break classical encryptio You got a source dor that?

>>22561 > 3) Cryptographers like to have *huge* security margins. AES-256 would take thousands of years to break with all the mass-energy of the earth converted into a perfectly efficient quantum computer. Is the situation actually that positive?

>>22601 Yes, it is. Some people have tried to calculate an "ideal computer"'s properties, I did the calculation a while ago. 126 bits of security is quite a lot.

>>22601 Idk if the other lain did their math right, but Grover’s algorithm runs in quantum time O(sqrt(N)) for black-box function inversion. So for AES 256 this would be (abusing notation) O(2^128). The earth, relativistically, has 10^40 Joules of energy and we can probably assume 6GFLOPS per watt. So from there you can finish the calc.

>>22601 >Is the situation actually that positive? Nta. It's theoretically possible that AES (or any other thought-to-be-secure symmetric encryption algorithm) could have some devestating vulnerability that we're just not aware of. But at this point in time, after all the research on AES and based on the way AES works, that seems unlikely. So barring some yet unknown vulnerability, in a post-quantum world, AES-256 would still be secure. Grover's algorithm would reduce bruteforce from O(n) to O(sqrt(n)). Cryptographic standards used to consider 128 bits of security as enough to be secure for forever, but in face of Grovers, a cipher with 128 bits of security against classical computers is reduced to 64 bits of security against quantum computers. So now we just use AES-256 and have 128 bits of security against quantum attacks. To put 128 bits of security into perspective, imagine a computer that ran at a trillion, 10^12, cycles per second (our fastest clock speed right now is about 9 billion cycles per second, roughly 1% of this imaginary trillion cycles per second cpu). Now imagine that a full cracking attempt only takes 1 cycle (this is obviously unrealistically optimistic). Now imagine you have a trillion of these processors all running in parallel with no overhead. It would still take, on average, almost 2 billion years to successfully crack one ciphertext. I guess another thing worth noting is that saying just "AES" is a little simplistic. The different modes of AES have different use cases and improper mode selection and use can have devastating impacts. ECB mode and CBC mode especially have issues and should be avoided unless the implementors really understand what they're doing.

>>22620 >on average, almost 2 billion years to successfully crack one ciphertext. I made a mistake: its 2 billion DAYS, not years. It's roughly 5 million years.

>>22546 >store now decrypt later threat model It already is in place since forever. When computers became more accessible (just after WW II) they brute forced all remaining enigma messages to decode them. This isn't a new thought

>>22546 >The feds could take your tapped data, decrypt it with quantum computers, and then analyze it at scale using neural nets. >Is it over privacybros? Of course not. Obviously not. I really wish users would put as much effort into research as they did into finding screenshots of schizophrenics before posting. Only one of you in this entire thread seems to actually know what he's talking about, and even he isn't hitting the heart of the matter (though his answer is still good enough): With gHz clock speeds, privacy will NEVER die if you don't let it, regardless of quantum fantasies. A little basic complexity theory: Firstly, if P<<NP, most all experts agree that BQP<NP, as well. Secondly, workable encryption exists even if P=NP. Therefore, security is always possible regardless of whatever quantum computers are used for; just use an NP-safe scheme and you're future-proof. Now, these algorithms aren't perfect, they take much, much longer to encrypt and decrypt, but today's consumer conputers decrypt today's standards faster than the spring under your enter key recoils so there's plenty of slack to take up. Maybe the whole thing takes two real world seconds, in which case it's done faster than the metric ton of javascript that will render your encrypted data takes to process, and maybe your computer also spends most of that time doing something else because you have a purpose-built encryption coprocessor on your motherboard like ye olde math coprocessors that gets all of the work farmed out so your system can multitask your 40-million line operating system with it. That's all the actual tech change you'd need to keep encryption alive literally forever, since P=/=EXP, and most of the work is already done, since they've been common features on mainframes for a long time to help business vendors brag about security to enterprise customers without knowing it themselves. An enterprising lainon just needs to implement the right algorithm in silicon (hopefully as public domain hardware so everyone jumps on board the same standard) and start a raspberry pi-like ecosystem to popularize expecting it and the day will be won against the glowies forever. No matter how many encrypted messages they archive or what they invent it will always be too expensive to bother trying to use them at any general scale.

If the de Broglie-Bohm pilot wave theory accurately explains quantum mechanics, then quantum computers would not function. According to Bell's Theorem, pilot waves would only have to be non-local.

>>22601 >Is the situation actually that positive? It's not that quantum computers will be faster and therefor brute force quicker, it's that they process data in a different way which makes new attacks possible. The biggest danger that we know about is shor's algorithm which would break old versions of SSL/TLS, which is the encryption used in websites and other things like the connections between Tor nodes. Modern versions of TLS have added something called forward secrecy to make sure that even if somebody knows the ciphertext and the private keys they cannot find the session key to decrypt the data. But if you did some illegal soykaf over https or Tor a long time ago, and all that encrypted traffic is in a database somewhere, and viable quantum computers arrive, then you could be in trouble. The other thing is grover's algorithm but >>22620 covers that well.

>>23415 >de Broglie-Bohm pilot wave predict quantum computers would not function 't Hooft's cellular automata theory predicts quantum computers would fair no different than classical computers. https://arxiv.org/pdf/1405.1548.pdf