Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ok Google: please publish your DKIM secret keys

blog.cryptographyengineering.com

Post reply on HN
Print view View original

321–330 of 492 posts

Re: Ok Google: please publish your DKIM secret keys

#321

Post by tptacek »

baobabKoodaa wrote: post #298
GoblinSlayer wrote: post #281

Earlier quoted context omitted.

Receipts can be signed with gnupg, it actually provides non-repudiation guarantee you want, DKIM doesn't provide that, it's only for journalists.

> Receipts can be signed with... In the real world it doesn't matter which cryptographic protocols are theoretically available for use. What matters is which protocols everyone else is using. For example, in the case of receipts for purchases on the web, literally everyone is using email. You will not be able to get amazon to sign a receipt with gnupg. If you want to embark on a path of convincing the world to move a…

Which is why, when email providers can do very simple things that improve privacy for all their members, even though email is irretrievably insecure, they should do so, as harm mitigation. Your mail spool will eventually get owned up. The least they can do is make it deniable.

Re: Ok Google: please publish your DKIM secret keys

#322

Post by tptacek »

josephg wrote: post #289
tptacek wrote: post #199

Earlier quoted context omitted.

(a) I wasn't aware of this. (b) Thanks for the link! (c) The IETF is such a shitshow for cryptography.

The IETF is an open organisation. Specs are written & ratified by the people who show up. And often that means folks from large companies. If you don’t like that status quo, get involved. They would love to have you @tptacek.

They would absolutely not love to have me, and I keep away from their mailing lists as much as a favor to them as to my own sanity. I'm not alone in the wilderness on this opinion that IETF process produces abysmal cryptography, by the way; Bernstein beat me here by 2 decades.

Re: Ok Google: please publish your DKIM secret keys

#323

Post by legolas2412 »

beagle3 wrote: post #297
mthoms wrote: post #176

Earlier quoted context omitted.

Current events prove otherwise. See Hunter Biden.

I have not seen a single mention of DKIM w.r.t to Hunter Biden. Did you? Was any evidence presented? I couldn’t find any. I fail to see how admissibility or lack of it, in a court of law or of public opinion, has anything to do with DKIM+Hunter Biden. Can you elaborate?

I saw this news (https://www.washingtonexaminer.com/news/cybersecurity-expert...) a few days before the election. There is also a github repo.

I am not sure why the DKIM for all emails were not released, or why this did not catch more media coverage by other news organizations I consider more reliable (like NYT).

Re: Ok Google: please publish your DKIM secret keys

#324

Post by jchook »

hpoe wrote: post #18

So the author's central thesis essentially seems to boil down to that leaked emails were able to be cryptographically verified, because of DKIM and so we should prevent that so people can't use email to blackmail politicians? Ultimately I prefer the more information that we can get on politicians available. It seems to me that especially when an elected official has something they don't want others to know about that…

DKIM was not designed to authenticate emails far into the future.

In fact, it does not authenticate any emails without a corresponding public key currently published to DNS. It provides specifically for "empty" or revoked keys to avoid such retro-validation.

Seems the central thesis is that because these messages are patently no longer authenticated by DKIM, we should eliminate any remaining hope of them being construed as authenticated by DKIM.

Re: Ok Google: please publish your DKIM secret keys

#325

Post by busterarm »

lxgr wrote: post #230
busterarm wrote: post #62

Earlier quoted context omitted.

Nobody is telling you to not be gay. If you being gay is a secret, then don't send that secret over _plain fucking text email_. Do you send your social security number to people in emails? Email has never been privileged communication and the problem isn't one of validation but one of not understanding one's level of privacy and risk. It uses relays without end-to-end encryption and there's no guarantee that what you…

If a global infrastructure provider can, without downside, offer non-repudiability – why shouldn't they? Not everyone is tech savvy. Not everyone understands encryption. Not everyone makes rational choices all the time. Does that mean everybody should suffer the consequences of an arguably unintentional side effect of the technical implementation of DKIM?

> Not everyone is law savvy. Not everyone understands legalese. Not everyone makes rational choices all the time.

> Does that mean everybody should suffer the consequences of an arguably unintentional side effect of the technical implementation of law?

Thankfully this is not law, but people should understand the things that can get them in serious trouble. A lot of legal concepts follow from basic principles and history.

Privacy concepts are the same way. It's simply not sufficient that we don't educate everyone about these things anymore. Ignorance isn't going to protect anyone from the fallout of misuse of technology.

The solution isn't to coddle people, it's to provide better technology that does the thing the way people intend to use it.

Re: Ok Google: please publish your DKIM secret keys

#326

Post by beagle3 »

legolas2412 wrote: post #323
beagle3 wrote: post #297

Earlier quoted context omitted.

I have not seen a single mention of DKIM w.r.t to Hunter Biden. Did you? Was any evidence presented? I couldn’t find any. I fail to see how admissibility or lack of it, in a court of law or of public opinion, has anything to do with DKIM+Hunter Biden. Can you elaborate?

I saw this news ( https://www.washingtonexaminer.com/news/cybersecurity-expert... ) a few days before the election. There is also a github repo. I am not sure why the DKIM for all emails were not released, or why this did not catch more media coverage by other news organizations I consider more reliable (like NYT).

Thank you for this link, this did not come across my radar.

From your link:

> The only way the email could have been faked is if someone hacked into Google's servers, found the private key, and used it to reverse engineer the email's DKIM signature, Graham, said.

https://www.zdnet.com/article/google-fixes-major-gmail-bug-s... is from Aug 2020 and discusses an SPF/DMARC vulnerability that was in Google since forever (and though reported 4 months before public disclosure, was fixed only 7 hours after public disclosure). The last google DKIM bug I'm aware of was in 2012, so I can't counter the specific claim about DKIM with evidence, but the assertion that "the only way to spoof x is to hack and get the private key" is not any absolute truth.

(P.S: I have seen no denial nor confirmation about the authenticity of the Hunter Biden data - only claims of Russian involvement. Make of that what you will. The DKIM is circumstantial data until there is confirmation or denial - especially, as you say, it's not all released).

Re: Ok Google: please publish your DKIM secret keys

#327

Post by __david__ »

speedgoose wrote: post #85

I doubt Google will publish old private keys that were not designed to become public later. I would guess that it's too dangerous or cumbersome to do the security analysis. What if someone realizes that Google uses a broken cryptographically secure pseudorandom number generator (CSPRNG) à la Debian ? Unlikely but the risks exists, so not going to happen in my opinion.

> What if someone realizes that Google uses a broken cryptographically secure pseudorandom number generator (CSPRNG) à la Debian ?

1. Someone with bad intentions figuring that out could start spamming other domains using gmail.com From Addresses.

2. Someone with good intentions would contact google security for a bug bounty or maybe just publish a zero-day report. Google would correct the issue and the world would be a slightly more secure place.

#2 would almost certainly happen, I suspect. And if #1 happened _before_ #2 then there'd be more spam in the world, temporarily.

To me the risk seems low.

Re: Ok Google: please publish your DKIM secret keys

#328

Post by couchand »

beagle3 wrote: post #301
woodruffw wrote: post #181

Earlier quoted context omitted.

I don't think it's a fascination, it's what the OP is about. We're talking about the subject of a blog post, no? I think the point boils down to expectation management: journalists (and ...) barely understand non-repudiation, much less why each of the following scenarios pans out: * 2006 email + 512-bit RSA, leaked in 2006: probably authentic * 2008 email + 512-bit RSA, leaked in 2012: potentially inauthentic * 2008…

The right solution in this case is to educate journalists - they are up to date on things like deep fakes and should be on DKIM. The wrong solution is to make previously private keys public to make any reasoning about past data impossible in the name of “hut journalists might get a wrong impression”

The OP is looking for a systemic solution to the problem, somewhat akin to the way establishing a bug bounty program aligns incentives. Your solution is like asking your team to please work harder to not release exploitable bugs.

Re: Ok Google: please publish your DKIM secret keys

#329

Post by legolas2412 »

beagle3 wrote: post #326
legolas2412 wrote: post #323

Earlier quoted context omitted.

I saw this news ( https://www.washingtonexaminer.com/news/cybersecurity-expert... ) a few days before the election. There is also a github repo. I am not sure why the DKIM for all emails were not released, or why this did not catch more media coverage by other news organizations I consider more reliable (like NYT).

Thank you for this link, this did not come across my radar. From your link: > The only way the email could have been faked is if someone hacked into Google's servers, found the private key, and used it to reverse engineer the email's DKIM signature, Graham, said. https://www.zdnet.com/article/google-fixes-major-gmail-bug-s... is from Aug 2020 and discusses an SPF/DMARC vulnerability that was in Google since forever (…

Sure, you raise very important points. I just found it weird that NYPost was happy just releasing the emails and not the DKIM, and when one was validated, it received literally no coverage. I thought it might catch steam after the election, but the literal silence is surprising to me.

I am not insinuating any wrongdoing from anyone, just bringing it to your attention, as you claimed to not know about it.

Re: Ok Google: please publish your DKIM secret keys

#330

Post by whoisburbansky »

couchand wrote: post #328
beagle3 wrote: post #301

Earlier quoted context omitted.

The right solution in this case is to educate journalists - they are up to date on things like deep fakes and should be on DKIM. The wrong solution is to make previously private keys public to make any reasoning about past data impossible in the name of “hut journalists might get a wrong impression”

The OP is looking for a systemic solution to the problem, somewhat akin to the way establishing a bug bounty program aligns incentives. Your solution is like asking your team to please work harder to not release exploitable bugs.

Agreed, grandparents proposed solution is basically the same as pushing for users to be better trained to protect themselves from phishing, which has shown to be ineffective time and again and is downright masochistic when an easier system solution exists.
Post reply on HN