Hacker News Reader

Comment by chenzhekl | Hacker News Reader

The title on HN is a bit misleading. You should use the original title of the article. According to the article, what requires an import permit is the "foreign 'data encryption technology employing a key length greater than 256 bits'". In reality, the encryption standards (SM1, SM2, SM3, SM4, etc) in China allows strong encryption. I don't see any clues stating that China forbids encryption technology with a key greater than 256 bits.

View on HN

> In reality, the encryption standards (SM1, SM2, SM3, SM4, etc) in China allows strong encryption.

Those are not strong encryption standards. They have dual key escrow such that the government can backdoor the connection with their own key. Those are China made encryption standards for use primarily with government products but they are attempting to also force them on the general public.

> I don't see any clues stating that China forbids encryption technology with a key greater than 256 bits.

That is exactly what this results in, because your choice is either a government backdoored encryption method (which basically means it isn't encrypted at all) or a non-backdoored encryption method that is limited to a key size no greater than 256 bits.

I will note, with this over reliance on dual key escrow. If a foreign government were to steal that key through espionage without China's knowledge, it would let say the US government to backdoor China's own government communications.

> I will note, with this over reliance on dual key escrow. If a foreign government were to steal that key through espionage without China's knowledge, it would let say the US government to backdoor China's own government communications.

Why would a government use key escrow for its own internal encryption (for actual important communications, as opposed to "retail" business functions like running garbage pickup operations)?

> Why would a government use key escrow for its own internal encryption (for actual important communications, as opposed to "retail" business functions like running garbage pickup operations)?

Because governments (especially dictatorships) are most scared of their own members who can possibly seize power.

Your question presumes a Western, liberal paradigm of how govts should function. Most of the answer rests in different levels of institutional mistrust. Authoritarian govts are necessarily always operating under a siege mentality, so they must check on each other. Noting that, at the highest level the Chinese govt is wholly controlled by the Chinese Communists and the Communists correctly note the existential threat their bureaucrats pose. And it cascades down from there. There’s an illustrative proverb that goes “Heaven is high and the emperor is far away.” This idea justifies doing something you know is wrong, but is unlikely to be punished.

So to answer the question, the govt is not using key escrow for its own internal communications. Because it isn’t a monolith, and it’s not internal. The different fiefs need to be watched as much as anyone else.

> According to the article, what requires an import permit is the "foreign 'data encryption technology employing a key length greater than 256 bits'". In reality, the encryption standards (SM1, SM2, SM3, SM4, etc) in China allows strong encryption. I don't see any clues stating that China forbids encryption technology with a key greater than 256 bits.

I guess that's a nit against the post title, but I don't think it really clarifies much.

So they have domestic >256bit ciphers, and this restricts foreign equipment that uses >256bit ciphers. China also has an extensive domestic surveillance system, has lots of weird "national security" regulations (e.g. they use an obfuscated coordinate system for public maps, and IIRC it's illegal to use a GPS receiver for anything resembling mapping), and is also pretty protectionist.

What's the purpose of these regulations? Is it...

1. To further strengthen domestic surveillance by encouraging the use of (possibly compromised) domestic encryption equipment, or...

2. protect domestic industry by making certain technology imports difficult, or encourage foreign entities to buy Chinese technology for interoperability reasons, or...

3. discourage the domestic use of foreign equipment on national security grounds (e.g. foreign backdoors), or use interoperability or market-access concerns to weaken foreign equipment, or...

4. all of the above?

I tentatively conclude from this news that they have a Quantum Computer that can break symmetric key encryption with 256 bit keys or lower, or hope to have one soon, and that they don't have to worry about this domestically because they have enough control over the endpoints.

On a side note, my own encryption software is now prohibited in China, but that hardly matters. I don't recall having any customer in China anyway.

iirc; Quantum computers do not make symmetric encryption weaker. They make asymmetric encryption relying on factoring primes weaker.

They effectively half the size of symmetric encryption keys (via Grover). They aren't considered as "breaking" them because doubling key size is pretty cheap. 256 bit symmetric is generally considered relatively safe vs quantum since a 128 bit key is still quite strong. That said, transitioning to 512 bits wouldn't be bad, since that would pretty much completely nullify quantum assisted breaks.

128-bit is still absolutely impractical to brute force now or in the foreseeable future. Even 96-bit is questionable, though likely achievable after some time by someone with a lot of money who is able to create a huge farm full of ASICs for the task.

A QC large and stable enough to run Grover's Algorithm would be a problem for symmetric keys and hashes smaller than about 192 bits. Most cryptographers recommend 256-bit or larger for a good margin of safety.

Asymmetric crypto is more complex story.

Furthermore, why brute-force crack crypto when you can inject code using the log4j exploit which transmits the private keys.

Sure log4j might have been recently patched, but it's not unrealistic to think that a nation-state has access to similar exploits.

Yes, a thousand times. The vast majority of compromises happen because of software exploits or social engineering.

This is not an argument for using shitty cryptography, but it is IMHO an argument for being more afraid of the implementation and the human beings using it than the crypto.

Does anyone know how this interacts with the AES-256 break from a while back? Something to do with, they didn't use enough rounds so the security is really only about equivalent to AES-128. And then nobody cared because having the security of AES-128 isn't a practical break.

Unless we get quantum computers. Then what happens with AES-256?

I haven't heard anything about not using enough rounds, so it's possible that my information is just out of date. My understanding is that AES-128 was chosen to be resistant against classical attacks, and AES-256 was chosen to achieve the same level of security as AES-128 against attacks that incorporate quantum computers and Grover's algorithm.

As I recall it was about the key schedule. But, same outcome. Some people, idiotically, use 256 bit AES anyway, because it reassures less literate customers. Sort of like the patdowns at airports: security theater.

They basically half the symmetric keyspace (but there could be better, more specialized quantum algorithms), and the next smallest keysize is usually 128 bit. So that's what most foreign encryption will use after this new law. A brute-force search of a 64 bit keyspace is no problem for a nation state like China.

Oh, please tell me that machine will break Bitcoin, so I won't have to listen to these religious crypto people anymore!

I didn't know China had their own encryption standards. Do American companies in China (Apple Facebook etc) use them?

RISC-V has some extension instructions dedicated to them. The extension was ratified very recently, so it isn't clear which chips will actually support them.

> Facebook

While FB have many customers in China (much as Google do), I don't believe that they have any actual data centres, and they certainly don't have any (official) users.

They're just horribly modified versions of western standards. And no they're only used by governments in China. The general public still uses TLS.

So... in other words, encryption with a key greater than 256 bits is forbidden unless you use one of the state-sanctioned (read: backdoored) algorithms. Neat.

I suspect this more the Chinese equivalent of America’s FIPS 140-2.

https://en.m.wikipedia.org/wiki/FIPS_140-2

No not really. FIPS specifies minimum encryption levels and forces certain things to not be used that are considered insecure. However it doesn't cap the maximum key length that is possible to be used.

Also FIPS 140-2 is outdated, please see FIPS 140-3.

https://en.wikipedia.org/wiki/FIPS_140-3

That is not a correct paraphrase of what chenzhekl wrote.

Perhaps there is a rule that prevents someone in China from implementing any algorithm they want to implement with any number of bits they want to use, but that rule would hardly be an "import control", which is what the linked article is about.

Or it may just be, that anyone (currently) caring about >256 bits of symmetric key is likely not knowledgeable enough to avoid snake-oil salesmen (usually are annoying, but in crypto are outright dangerous).

Is there any well-defined, well-implemented cipher with a keysize of >256. AFAIR the AES is either 128 or 256.

That seems more (rightfully) careful than draconian..

The “key size” for aes-xts a mode used for disk encryption uses a 512 bit key. Effectively it is really 2 different 256 but keys though.

Ok, we've changed the title back to the article original - except with s/MOFCOM/China/.

Submitted title was "China forbids data encryption using a key greater than 256 bits". Submitters: please follow the site guidelines, which ask "Please use the original title, unless it is misleading or linkbait; don't editorialize."

https://news.ycombinator.com/newsguidelines.html

If you want to say what you think is important about an article, that's fine, but do it by adding a comment to the thread. Then your view will be on a level playing field with everyone else's: https://hn.algolia.com/?dateRange=all&page=0&prefix=false&so...

Please don't do this. The original title was correct.

The original title is "MOFCOM Issues New Encryption Import Control Effective Immediately". It's pretty standard to replace the name of a government agency with the name of the government or country, especially if the former is an obscure name that most of the audience won't recognize.

Reply on news.ycombinator.com