A planned EU regulation about website certificates is causing concern | Hacker News Reader

A planned EU regulation about website certificates is causing concern

feistyduck.com

156 pointshannob145 comments

This title is clickbait. The proposed amendment to the framework is just excessively bureaucratic around the management of cert authorities. This might potentially result in browsers accepting certificates from authorities with a slightly dubious track record (according to browser vendors), yes, but the certificates themselves will not necessarily be "less secure" - which the layman typically interprets as "cryptographically weaker".

In practice, this would simply move a little bit of control over the list of accepted authorities from browser vendors to EU institutions, which is not terribly unreasonable a stance.

In fact, as time passes and authorities become more and more tech savvy, I expect this sort of thing to happen more and more. There is a lot of unaccountable authority (eh) in internet infrastructure, with special companies like Google and Microsoft (and yes, Mozilla too) effectively holding the keys to many kingdoms. Even if such power is used scarcely and wisely, there is a question of legitimacy.

> In practice, this would simply move a little bit of control over the list of accepted authorities from browser vendors to EU institutions, which is not terribly unreasonable a stance.

And what happens when one of those institutions violates existing widespread CA policy, such as by issuing a certificate for MITM purposes, or otherwise making the web less secure? Right now, browsers can simply remove such CAs based on their established policy. (This isn't a hypothetical; browsers have had to remove CAs for such misbehavior in the past.)

Probably browsers will remove it anyway and deal with EU in court if they make a stink about it.

As it is, browser companies[0] are the sole authority on what is and isn't a valid CA. At least the EU leadership is elected.

[0] and let's be realistic here, this means Google.

> and let's be realistic here, this means Google.

Actually, it doesn't. There are four main browser CA root stores: Apple, Microsoft, Mozilla, and Google, and these are also the root stores of their respective operating systems (if you're on Linux, you're using Mozilla's root store, because distros do not want to run anything themselves).

Furthermore, the Microsoft, Google, and Mozilla root stores all also require inclusion in the common CA database, which is run by Mozilla. Also, Mozilla (and maybe Google) require CAs to participate in Mozilla's public lists, especially when it comes to notifying people of big CA security oopsies--all of the big CA issues have come to light on that list, and the coordination of the responses by browsers has been done by Mozilla.

So the biggest authority on who a valid CA is? Mozilla.

So, what you're saying is that Mozilla could sanction Russian CAs such that nobody would trust them?

First there would have to be Russian CAs. There aren't any. Most likely because Russians wouldn't trust a Russian CA, so what's the point setting one up?

If against the odds somebody had built a trustworthy Russian CA and it was currently in the root trust store for Firefox (and so likely everything else) then we might have this question to ponder, although the answer is likely "No" since this hypothetical CA isn't doing anything wrong, but that's a hypothetical very different from reality.

In principle, they could. In practice, you need a really good reason to do so. And "government-owned" isn't even sufficiently good reason, although "government-mandated MITM" is (see the banning of some Kazakhstan CAs for that reason).

No because that trust in Mozilla is only placed there because they only care about CAs operating properly. Nothing else.

They theoretically could, but why?

When it comes to which CAs are trusted, Mozzila is generally in the lead.

Not an elected group, true. But a foundation with a public charitable goal, and an entity with much less interest in being able to intercept communications than states.

I think it's better of in the hands of an un-elected foundation than in the hands of an elected, but politicized entity with interests in intercepting communication.

> When it comes to which CAs are trusted, Mozzila is generally in the lead.

Mozilla might well not be around forever, considering the state of their market share. And let's be honest, as long as their main source of funds comes from Google, politically speaking they end up as a fig's leaf around Google's interests. As far as we know, they might take certain decisions about the life or death of any given CA after someone at Alphabet picks up the phone. Maybe they don't do that, but there is no guarantee nor formal process to address this - particularly not in Europe, since all these are US entities.

I've watched how Mozilla's CA process operates in past incidents, as well as in the review of new CAs, and the criteria they use and the evaluation they do seems quite transparent.

> Probably browsers will remove it anyway and deal with EU in court if they make a stink about it.

A more direct and confrontational approach would be for browser companies to do what website owners have effectively done, which is to agree in advance to implement some sort of annoying pop-up (shown every time the browser updates) saying "Do you want the EU to weaken your security and enable spying? [Yes] [No]", like those annoying cookie pop-ups.

The difference being, of course, that the cookie pop-ups are designed to dark-pattern users into giving away their rights, and to blame the EU unfairly, whereas these browser pop-ups would be to help protect users' rights and to correctly blame the EU (assuming the regulation passes).

> At least the EU leadership is elected.

it isn't (directly anyway)

for the last set of elections: they even held televised political debates with candidates

then selected someone that hadn't even been mentioned (Von der Leyen)

By that argument, the German government isn’t elected either. The Bundestag is elected and elects a Chancellor who then picks a government.

But that very same argument, the use of an electoral collage of course also means the US president is not elected, since people's vote only choose electors.

how do you think the American people would feel if the college elected someone that:

    - hadn't won any primaries
    - hadn't attended any debates
    - said zero words about their plans for office
    - and hadn't even run for president

because that's exactly what happened with Von der Leyen's appointment

If Von der Leyen had even a third of the power Macron or Sholtz have, that could be an issue. I'd the French/German finance minister have more power than Von der Leyen. And once upon a time, the French Finance minister was an unelected banker, who never won a single election and did not work in the political field.

You’re arguing why the office doesn’t need to be democratically chosen.

That’s fair. A well run society has more than democracy - it also has skilled civil servants, a commitment to rule of law, protection for individual rights, etc.

But when people complain the EU isn’t democratic enough, they mean it’s too focused on skilled administrators.

That's because nation-state politicians are unwilling to give away powers, which is what happens when you focus away from skilled administrators.

the existence of other people in other countries doesn't mean the EU is fit to be called "democratic"

The existence of unelected political appointees does not make the EU non-democratic.

Furthermore, fixing the anti-democratic features of the EU requires giving it more power[0]. The reason why the EU has so many appointees is because it represents member states, not people. This is true for every international organization. The only way to have international organizations that are democratically accountable is to actually have a notion of international citizenship - in other words, to agree that, say, German votes are allowed to dilute Spanish ones. This is tantamount to political union.

Because the EU did not want a political union[1], they instead decided to limit the amount of elected representatives in the system, and let member states fill in the rest. So you don't get to vote for Ursula von der Leyen, you get to vote for the person who appointed the person who gets to vote for Ursula von der Leyen. Yes, this is complicated and stupid, but it's also the thing that keeps the EU international rather than sovereign. If you don't like it you should push for reforming the EU into a properly democratic single country.

Also, if you think the EU is entirely undemocratic, then let me ask you: when was the last time you voted for anything in the World Trade Organization, NATO, or the United Nations? The WTO in particular are the kings of antidemocratic policy laundering, so much so that it would make Ursula von der Leyen blush.

[0] For the record, giving it less power reverts back to the international default, where governments are entirely sovereign within their own territory and powerless without. So, say, the French have 100% say over who their leaders are and what they do, but 0% say over international affairs, because they aren't citizens of those other countries.

[1] This is almost certainly no longer the case now that Britain is out of the picture.

> The existence of unelected political appointees does not make the EU non-democratic.

not alone, no, but when combined with:

  - a parliament made up of directly elected representatives that can't legislate, modify or repeal previously passed acts
  - the ratcheting of transfer of powers in direct opposition to what the people of Europe want (ignoring No referendums on the Constitution of Europe and passing it through regardless): the "Monnet method"
  - deliberate "degenerative proportionality": a Maltese voter has 30x the representation of a German (makes no sense in the organ that's supposed to represent the people)
  - the concentration of executive power via the mass transfer of power from elected national parliaments to the council

it's beyond a joke to say it's democratic

> If you don't like it you should push for reforming the EU into a properly democratic single country.

I don't have a problem with the idea of a federated Europe, as long as the people voted for it

the problem the EU has is: the peoples of Europe wouldn't, so it's being done by slowly by deception instead

and by design it's impossible to revert (unless you leave entirely, which is extremely difficult)

> Also, if you think the EU is entirely undemocratic, then let me ask you: when was the last time you voted for anything in the World Trade Organization, NATO, or the United Nations?

I don't remember these organisations pretending to be democratic (let alone having legions of fans online stating that they are)

>I don't remember these organisations pretending to be democratic (let alone having legions of fans online stating that they are)

Does it matter? They're still a bigger stumbling block to democracy than the EU's various collections of appointed positions are. All of the anti-democratic features you mentioned are present in other international bodies, but amped up to 11. It is unfair to single out the EU as wholly responsible for this.

Also, I feel the need to defend the EU on one other point: ratcheting. If you have a free trade area in which member nations are allowed to pick-and-choose each little bit of the EU treaties, then you lose the benefit of the EU, and just get a bunch of top-heavy bureaucracy with no other benefit. Furthermore, the things that the EU tends to be really adamant about - such as freedom-of-movement[0] - are essential for making the bloc work at all.

The EU has such dogged fans because people don't want things like freedom-of-movement taken away from them, and nobody's made a serious proposal for fixing the EU's problems aside from "leave and go back to having borders everywhere" or "federal Europe". Alternatively, they're Americans who happen to oppose strict immigration control[1] and, thus, oppose euroskeptics purely by association.

[0] This is also why the UK's original "free trade without freedom of movement" proposal was shot down so quickly. The biggest mistake the west made was convincing the whole world to let goods move faster than people do. This constitutes a transfer of power from labor to capital.

[1] As you might have already guessed from the previous footnote I am an American who opposes strict immigration control. I'll at least entertain the thought of the EU needing reforms, though.

> Probably browsers will remove it anyway and deal with EU in court if they make a stink about it.

That doesn't sound like a sad state of affairs to you? It's like responding to the PATRIOT act by saying "Probably tech companies will refuse to comply and deal with the federal government in court if they make a stink about it."

That sounds like a great state of affairs, if people actually raise challenges to such laws, and win. I would love to see such overreaching attempts tied up endlessly in litigation, ideally with associated PR to make sure that the visible actions are all portrayed as a mess associated with the officials supporting such overreach in the first place.

I'd rather see overreach stopped before it starts, but in the absence of that, I'd like to see it stopped by any means we have, including civil disobedience and litigation.

In this context: remove CAs that issue MITM certificates, and loudly and repeatedly say "So you're prosecuting us and demanding that we include a known-insecure CA that's used to make the Internet less secure? Is that what you'd like to say to these cameras, the entire Internet, and people who will see this played repeatedly while they decide whether to vote for you in the next election? Or would you like to drop this?"

I mean, what's the alternative? We see people here complain every single day about unilateral action taken by Google and the like. Which is better, to have to deal with the rule of laws written by an elected body in a court, or to shout into the wind at a corporation?

>We see people here complain every single day about unilateral action taken by Google and the like.

While that's true to some extent, I don't find that to be the case in this specific area (ie. trust store and/or encryption). Are the browser vendors unfairly applying rules (ie. favoring incumbents)? Are their requirements too arduous? Are they not revoking bad CAs? It's not clear what this regulation is supposed to solve.

Third option: work with a transparent Open Source organization that processes CAs on the basis of keeping the Internet secure, where nation-state-level interception is considered part of the threat model.

Historically google and apple has followed mozillas decisions on CAs most of the time (google didn't even have their own program for a long time). I think it is still somewhat accepted that if you get approved by mozilla you will probably get approval from apple and google pretty soon. MS can be a bit weird though IIRC.

In practice there is every reason to believe it means m.d.s.policy, and thus anybody who cares.

In principle Apple, Google, and Microsoft can do whatever they want, but well, go back and see the history of such interventions for yourself and draw your own conclusions about where the crucial decision was made.

> Probably browsers will remove it anyway and deal with EU in court if they make a stink about it.

So how about we stop this proposal in its track right now instead?

Why? If given the choice between allowing corporate control of something and allowing an elected government control of something, it is not immediately obvious to me that leaving everything in corporate hands is a good idea.

You're right, it's not immediately obvious across the board. But in this very specific case, Mozilla is making an argument that the government standards are meaningfully less secure than their own.

Involving a government doesn't magically make things better 100% of the time, the actual tangible proposal still has to be better than what is happening in the market. Is that the case here? We don't need to have an abstract argument about whether governments or private corporations are always better; look at the specifics of this proposal and see if they're good for browser security.

We have a lot of data that suggests that EV certificates do not provide meaningful security because the "green lock" doesn't actually guarantee what consumers think it does. Independent of what you think about governments overall, is it good to have a new standard that ignores that research in favor of having yet another kind of special certificate that is displayed to users as more trustworthy?

In this specific case, Camerfirma is still considered by the EU government to be a trustworthy certificate provider even though all major browsers have blocked the company for repeated mistakes/policy violations. Independent of whether or not you trust the government overall, is a standard that would require browsers to trust Camerfirma a good idea?

I don't think we should immediately assume any government involvement in the market is a bad idea, but there's also real danger in saying, "well, in theory it might be preferable for this to be more democratic", and then never asking the followup question, "but in practice, is this specific proposal good?" Yes, you're 100% correct, it's not immediately obviously bad across the board for the government to control something. But when I go beyond what is immediately obvious to me, then I start seeing arguments from Mozilla and other security researchers that (in my mind) make a lot of sense.

EU isn't an elected government - the Parliament has no real power. EU is actually a dictatorship of bureaucrats.

The current way TLS is managed has far more democratic legitimacy. You can pick your browser and operating system i.e. root store, and then edit it if you want. EU dictating one is worse in every possible respect.

Exactly this. Trust must remain configurable by many instead of by a single authority.

Well, because as you can see, this law is already introducing strife where there should be none, by unskillfully requiring something that's not useful. What problem is this supposed to be solving again?

By rolling out QWAC, we would be reducing the agility of the ecosystem and preventing automated certificate issuance methods like LetsEncrypt which have improved the status quo immensely. And for what gain?

This is not a debate over whether certificate issuance should be controlled by corporations versus the government.

> In practice, this would simply move a little bit of control over the list of accepted authorities from browser vendors to EU institutions, which is not terribly unreasonable a stance.

I have zero trust in the competence and intentions of EU institutions

I already have to trust the firm that maintains the browser, how does adding another party help exactly?

this proposal's endgame is mandatory key escrow (attempt number #8 or so)

> I have zero trust in EU institutions,

Why is that?

Browser vendors have an incentive to maintain trust, it's their stock in trade. If they screw up, it directly affects their bottom line because people will move elsewhere. Their goals align closely with my own.

A foreign government (or even my own) has somewhere between zero and weak alignment with my goals. There's little reason they would want to choose good certificate authorities and strong reasons why they would choose ones their intelligence agencies can backdoor. Why should I add them to the chain of trust?

> it's their stock in trade

Their stock in trade is being easier to use and more feature-rich than the competition. Trust doesn't factor into it, in the sense that trust in a browser does not come from track record: It's all marketing, all the way down. Thus, if they screw up it does not matter. This point is easy to see because they frequently do screw up and it hasn't had any impact so far. There's no reason to assume it will have one some day.

Where exactly do people move if they only chose between Firefox and Chrome, there is not enough competition in the browser market

Browser vendors monetize your data, that is their main trade. Having lots of users doesn't matter if you don't control them in some way, such as decide who gets to be trusted as a CA. The browsers are supporting the current tech hegemony for CA's, for some reason Americans are much more "trustworthy" than people from anywhere else, do you really believe that? Why should the people in Europe be forced to trust shady American companies like Google or Microsoft who are or likely will get compromised by the American state? It makes sense for them to regulate this so that they can have European options.

Because of people like Axel Voss, Von der Leyen etc.

https://qz.com/1389385/article-11-and-article-13-axel-voss-i...

Because they have a demonstrated track record of incompetence combined with micromanagement combined with epic levels of corruption.

I have zero trust in the intentions of browsers, and at least a little bit in the EU. Now what do we do?

The problem is that you have to trust the browser either way.

Nothing, so we keep the status quo?

the status quo favours one of the two commenters upthread, so it makes sense that the other side will keep pushing.

Compromise will be required.

There are some people who think women should not be allowed to vote. Or that pi equals 3. Just because different opinions exist does not mean compromise is required. Sometimes it's perfectly okay to just tell someone to fuck off.

Telling people to fuck off when they argue for discrimination is fine, but doing so when they want accountability is the definition of tyrannical behaviour. Tyrannies, even enlightened ones, eventually tend to fall.

The current system has done a very good job keeping certificate authorities honest and secure by holding them accountable. Those who disagree can easily modify the trusted root certificates on their own devices to their liking.

The article presents several arguments why the proposed legislation would be worse than what we have now. Perhaps those who support these plans could try countering these arguments, rather than trying to force a "compromise" while accusing the other side of tyrannical behavior.

Some might argue that the government enforcing which websites are trusted looks a lot like a tyranny.

> Perhaps those who support these plans could try countering these arguments

These discussions are being had, as part of the EU process. This is a proposal, not a decree, and is being submitted by a body with some democratic legitimacy.

> Some might argue that the government enforcing which websites are trusted looks a lot like a tyranny.

As long as the government is an actual expression of popular will, mediated through the various mechanisms of social democracy, it's hard to argue that it would be a less democratic stance than letting an unaccountable foreign company in charge of things.

> but the certificates themselves will not necessarily be "less secure"

I don't trust the EU to handle encryption. I don't think that is unreasonable especially with the attempts against it cloaked as "reasonable access".

The chain of trust has weaknesses and any government institution can only fully undermine that trust. It would not last a week without abuse, the same happened with Covid tracking apps and location information.

> there is a question of legitimacy.

For the EU there is actually quite a lot of that. It is true that at some point we have to trust someone, but which problem do you want to see solved here?

I actually have move trust into the EU institutions than into the three main browser vendors. The EU is certainly more legitimate in most respects than various US private organisations.

What do you expect the E.U. to do that Google or Microsoft isn't doing already?

I don't generally use browsers of these companies, but if they break the chain of trust they would damage themselves as much as any user. I believe in the self interest of these companies and the understanding how the effectiveness of TLS is reliant on trust. I would trust Google more than Microsoft that likes compliance a little bit too much, but since both want their browsers established, any compromise to security would disqualify them immediately. The EU has ambitions for surveillance as far as I know.

They're already breaking the chain of trust. One of the EU's TSPs was dis-trusted by Mozilla after a long history of misissuance and BR violations. This isn't hypothetical; the EU is mandating the inclusion of insecure CAs.

I expect them to enforce slow, ponderous change processes which make it impractical to remove any but the most egregious violators from the trusted CA list.

> the same happened with Covid tracking apps and location information.

Do you have (a link to) more information on that? As far as I have seen (in NL) the Covid tracking app was designed and implemented fairly well and abuse has been very limited to non-existant.

The Luca-App in Germany has been misused DESPITE legislation that reserves use exclusively for Covid-Tracking. Health authorities have been complicit in this. Afaik there where no repercussions for the offenders.

The Luca-App drives me crazy. A private app pushed by some local governments, while an official, public one was available. Just why?

The Luca App is a privately developed app that a few german states paid for. The Corona-Warn-App has had no issues with privacy as it was designed privacy first and there is no possibility of data abuse to my knowledge.

Here is an English source I could find:

https://digit.site36.net/2022/01/22/tracking-corona-infectio...

There are two prominent apps in Germany. The other one is indeed secure against such attempts because it is technically not possible to get the information. I wouldn't even accuse the investigators here, it just shows that if there is a possibility for tracking, that it will be used. For better or worse reasons.

>I wouldn't even accuse the investigators here

There is a legal mandate which forbids use other than for Covid-Tracking. There is no excuse for anyone involved in this.

> in NL

The Dutch app was full of security holes.

> abuse has been very limited to non-existant

As far as you can tell, which is not at all. That begs the question, why do you trust that this is the case?

> The Dutch app was full of security holes.

Source? I tried a quick search and haven't found anything.

> I don't trust the EU to handle encryption.

Literally nowhere in the article it says that.

The TL;DR is: "QWAC certificates will have to show the green banner" and maybe non-QWAC will have some kind of warning.

The link on TFA is informative https://scotthelme.co.uk/looks-like-a-duck-swims-like-a-duck... (sounds like a CA lobbying for this crap)

In the end I suspect browsers might revoke some QWAC certs if they feel they are invalid

The whole article is about mandatory certificates, of course that is basically taking control of the most widespread encryption in use today. Encryption that has its weakness in the chain of trust, which the EU wants to have an influence on.

So yes, this is essentially handling encryption if the EU indeed wants to enforce mandated certificates. If you want to discuss technicalities and the problem of key exchange we can do that too.

Does the EU government, or the US government, or any other government for that matter strike you as doing a particularly good job of legislating around the internet?

Personally, my impression is the legislation is at best hit-and-miss. You might get some great privacy protection. You might get mandatory encryption back doors. You might get free mobile data roaming. You might get cookie warning messages. You might get worldwide license-free spectrum, some of the most economically productive spectrum ever. You might get a patent system that lets people patent slide-to-unlock. You might get the state funding university research and giving the results away for free. You might get the DMCA giving youtube creators getting copyright strikes because someone's copyrighted silence. You might get cable rights of way and common carrier protections, which are vital to the communications industry working. You might get government ID checks for social media users.

Seems to me legislation is a coin flip - sometimes great, other times completely boneheaded.

I trust the people I implicitly have to trust (browsers could backdoor your browser, so why would they trust a dodgy certificate). I don’t trust a government, especially not my own, to handle certificate authority selection without involving politics.

Hey, nothing keeps you from managing your trusted certificates on your machine, right? Maybe this initiative can even highlight some issues regarding dubious CAs and make people care about whom they trust more explicitly.

Ah yes, I’ll just tell grandma she should just validate the corporate behaviours of certificate authorities herself. How did I not think of this before?

CAs with a dubious track record are a far bigger problem in terms of practical security than weak cryptographic keys.

Both are equally important because either one will break the security. The CA problem is the harder human problem to solve though.

With the commodification of free and automated certificates, I'd argue that authorities become less and less tech savvy: There is little to no money in the business sector, which means there's less possibility to keep technical people around.

Also, I am surprised what exactly is making you think that the baseline requirements do not exist for security reasons (even if it isn't exclusively about cryptographic security).

> With the commodification of free and automated certificates, I'd argue that authorities become less and less tech savvy

That's not what I meant. The point is that, as time passes, more and more people are involved in political decision-making who understand how tech works.

> what exactly is making you think that the baseline requirements do not exist for security reasons

Once you move out of the nitty-gritty of crypto tech, "security" is in the eye of the beholder. Is a solution more "secure", for society at large, if it allows terrorists to communicate undisturbed? You might say yes, others might say no. There isn't an absolute scale there. Choosing trust is one of those areas where the solution is political, not technical, hence forever debatable. So a statement implying that one set of compromises is "less secure" just because it moves trust from certain subjects to other ones (more formally accountable, in this case), is simply not objectively true, and really meant to muddle the waters.

Less secure is exactly that. If `Mr Shady` is in the mandated list (by law), and the browsers are required to threat their certificates as trusted, what do you think is the impact on the whole ecosystem?

I would be perfectly happy with MR shady issuing such certificates if he has €100 billion in escriw with EU authorities and is legally forced to cover all losses and fraud caused by his certificates. Then, perhaps, mr. Shady will think twice

>but the certificates themselves will not necessarily be "less secure" - which the layman typically interprets as "cryptographically weaker".

Less secure is more about trust than the crypto algorithm...

The strongest algorithm with no trust in the authority issuing it, is useless.

> but the certificates themselves will not necessarily be "less secure" - which the layman typically interprets as "cryptographically weaker".

I think this is word-games. If all we cared about was cryptographic security we wouldn't have have certificate authorities in the first place, making certificates that are themselves secure isn't hard. The only reason we care about certificate authorities is because of the security of the surrounding process: which is exactly the stuff that Mozilla is complaining about.

You're minimizing the most important part of a certificate authority. It's like going into a discussion about a potential phishing vulnerability and saying, "there's not security risk because the browser requests are still going over SSL."

The point of a certificate authority isn't just to have cryptographically secure certificates, it's to have a secure issuer. So it feels really misleading to completely jump over the concerns Mozilla has laid out about weaker standards for issuer security and say that it's fine just because everything is still encrypted.

What authority does the EU have to tell browsers what settings to use in their non-commercial product? The EU is always free to fork browsers or maintain a version tuned to EU standards and let users choose what they want.

The only thing worse the corporate control over CA's is government control over CA's...

There is alternative to both, the solution to corporations having too much power is NEVER giving that power to governments

Governments should have absolutely zero control or influence over the trusted CA list that browsers ship.

Hoping the browser vendors will enable this only for EU users, I want nothing to do with bad CAs.

> the certificates themselves will not necessarily be "less secure" - which the layman typically interprets as "cryptographically weaker"

We don't just need _cryptographic_ strength to deliver security, otherwise we'd all be using self-signed certificates...

How is the state of certificate authorities included in browsers these days?

I remember some years back there being much noise about supposedly atrocious behaviour by some of the default included cert issuers...

It's much better. There are processes to enforce good behavior, these are very transparent (basically everything is happening on public mailing lists, the mozilla bugtracker and wiki), and in extreme cases CAs get distrusted for real.

None of that is happening for QWACs.

> In practice, this would simply move a little bit of control over the list of accepted authorities from browser vendors to EU institutions, which is not terribly unreasonable a stance.

I actually do not think that is what we want at all.

I’m always amazed at HN posters undying trust of government with respect to technology which has repeatedly shown itself to be corrupt, inept and has never taken the time to think through unintended consequences.

Because private companies are always competent, never abuse their privileges, and always act in the best interest of society at large. /S

If I understand correctly, this legislation isn't about the government issuing certificates; it's about the government granting inclusion privileges to the same kinds of private companies that you're worried about.

I distrust private companies too, but I trust Mozilla more than I trust companies like Camerfirma, and we have a pretty long track-record that shows us that governments are typically slow to address misconduct in private companies they're partnered with.

I'm not sure I understand why this is being treated as a private company vs government debate; in my mind this is a debate over whether we should all be required by the government to trust private companies that Mozilla claims already today don't meet their security standards. It feels odd to claim that a government telling me I have to trust a private company fixes corporate incompetence or malfeasance. You don't see a potential for abuse of corporate privilege or for exploitative behavior when a private, for-profit CA can't be distrusted in browsers without them first petitioning the government?

The difference is that private companies don’t have the power to force me to do anything - the government can forcible take away my property and liberty.

The difference is that governments have a monopoly on violence. Corporations don't.

The problem isn't merely the fact that the EU wants some control over the process, it's the fact that the EU wants to revive a bunch of dead ideas and bad CAs in the process of doing so. The article specifically cites a CA that was banned from root trust stores which the EU wants to forcibly un-ban, which is very bad.

Yes - strictly speaking, those bad CAs can still issue "cryptographically secure" certs. However, the layman does not have a clue what "cryptographic security" means. They just hear "~~b̷̧͚́͝ȧ̷̠̜̦̉k̵̯̘̠͊̂e̴̞̓̃m̵̹͒̂͠o̵͎̣̎̓n̴͔̤̖̊̑o̴̙̰̼̍c̶͖̅͊h̶͔͛̇̓i̸͚̎n̶̮̬̊̒c̴̮̈̌ḣ̸̹i̵̤̊̈́n̴̹̮̾̃̈́~~ security", parse out the bits of English they do understand, and treat the rest as a linguistic intensifier, which is highly misleading.

In fact, "cryptographic" in this context specifically ignores the authentication of the key. It just means that it's computationally infeasible[0] to decrypt the message if you only have the ciphertext. However, in order to get this security, you need to make sure only the intended recipient has the key. This not only means that the recipient needs to keep their key safe, but that you need to make sure that you're using the intended recipient's key. Otherwise, an attacker could just hand you their key instead of the recipients, and then decrypt your messages and do whatever they want with them.

In other words, secure key authentication and exchange is a necessary precondition for cryptographic concealment.

In some simple cryptosystems, we don't talk about this, mostly because we imagine the recipient and sender exchanging keys in such a way that the system doesn't need to care about it. Perhaps they exchanged paper keys in person in the middle of a potato field in Idaho. However, this is highly restrictive; imagine having to drive to every tech start-up's HQ to copy some QR codes with their public keys on it before you could visit a new website. What we would rather have is a way to take one securely-exchanged key we already have, and use it to securely exchange more keys over the Internet.

That's what a CA ultimately is - an entity we already have keys for, that is capable of making sure our recipients are who they say they are. So if I go to news.ycombinator.com, the server I connect to will send me a message from that CA saying that they checked the server and validated that it's run by the person who owns news.ycombinator.com.

Except now that makes the CA the weakest link in the system. A CA that is malicious can totally strip any and all of the protections that HTTPS is supposed to provide. So browser vendors are really, really picky as to what CAs they put in their trust stores. CAs that habitually make mistakes or lie out of their asses get removed because they are not trust-worthy. Any legal proposal that would make it easier or even possible for them to reverse such a decision needs to be treated like cancer.

Yes, I know that browser vendors hold a lot of power in this equation - but there is little evidence that the measures they impose upon CAs are excessive or pretextual. Furthermore, the proposed replacement is making mistakes that the browser vendors aren't, so it's a strict security downgrade for no added freedom[1].

[0] trying every possible key requires more compute time than exists in the current universe

[1] CAs are sideloadable on every device ever. Yes, even iOS. However, the only use for this is for...

- Securing local web servers without an FQDN

- Using enterprise security/spyware/filtering products that need to strip HTTPS in order to work

I know of no case in which a CA has gotten into the business by telling people to install their alternative certificate. There are some edge cases in which very large CAs - specifically Let's Encrypt - have transitioned from having another root cross-sign them to becoming a root CA themselves. However, that only affected really old devices, and you could at least theoretically install the ISRG Root X1 cert on that device to work around this.

https://letsencrypt.org/docs/dst-root-ca-x3-expiration-septe...

CAcert.org tried years ago. LetsEncrypt has replaced the need for this service.

There was a proposed Kazakhstan root cert for nationwide MiTM, so you could argue that was its purpose of their CA. Darkmatter (which had somewhat of a legit claim as they are the local gov IT provider) tried to get root CA inclusion too, but the risk of MiTM was too great and was rejected -- allegedly, potentially, for this purpose too.

>certificates from authorities with a slightly dubious track record

CAs can sign certs for any website, that absolutely makes the web less secure.

One of the most relevant problems in Germany is that there is a caste that entirely specializes in written warnings and cease-and-desist declarations. A lot of our parliament consists of lawyers that tend to make favorable laws for their own industry. Just as background information on how Germany tends to enforce laws through private enterprise.

Really good behavior from the browser vendors. Either the TLS chain is working or it is not. There is little nuance in the chain of trust.

Since encryption seems to be the enemy dejure of governments, I would expect the next step would allow decryption by state executive forces if such "certificates" were established. This seems like getting a foot in the door.

But I still wonder how it would enforce compliance. Certificates can be added and removed from browsers. Some use the OS certificate store and some have their own. I hope the community keeps a watchful eye. Because a mandated certificate is the opposite of secure.

Belgium is very similar in that regard. Our politicians are very talented in turning what should be straightforward processes into complex bureaucratic puzzles that require at the least an advanced degree in accounting to solve. So frustrating as it's holding SMEs back.

I hope someone from the Chrome team reads this and decides to implement a separate certificate store like Firefox has.

Sorry for being a grammar nazi, but "enemy dejure" doesn't mean what you think it means - technically that form states that encryption is an enemy in law (de jure, in Latin - typically used in opposition to de facto, "in practice").

I expect you meant to write "du jour", French for "current, of the day (jour)". Fire your autocorrector.

It occurs to me, I'd be interested in my browser tracking the root certificates it uses and has encountered, and present me with a report of them. I've been interested in curating my cert list personally before, but without an idea of what certs I am in practice trusting (and not just in theory), it's really difficult to know where to start. You're presented with a list of a couple hundred or so root authorities with little more than their name to go by. I could get a real jumpstart if there was 80-90% that I simply never encounter and could easily see that.

One way browsers could push back at least a little on this would be to make cert auditing like this easier, and, if not already possible, potentially expose it to extensions. Preferably with a brand new "manage certificates" extension, which the browsers are welcome to and/or invited to festoon with whatever dire warnings about security they like. (It's hard, if not impossible, to create an interface to improve security like this that can't also be used to tear it down.) Giving enough visibility into the cert process that an extension could flag "hey, this is a new root cert you've never seen before" would be nice too.

Obviously, my metaphorical grandmother isn't going to be interested in this, but it could at least start a more community-oriented "conversation", in the form of running code, around what cert authorities I trust, rather than the browser coders.

I'm not looking to see if this already exists, because if it does I'm sure someone will know exactly where it is and what it's called and can link it to me, and everyone else reading. Please do, if it already exists.

The firm Camerfirma is provided as the poster child of why the EU can’t be trusted with the QWAC issuance rights - because it (EU) has been shown evidence of malpractice on the part of Camerfirma and failed to revoke its QWAC CA status. But looking at the page linked by the article, it seems Camerfirma’s QWAC CA status has in fact been withdrawn? The article doesn’t make this distinction clear. Or do I misunderstand?

I'm curious what causes this proposal to get attention now. As far as I can find, it has been submitted by the European Commission in June of last year, but little has happened since; the European Parliament and European Council in particular don't seem to have considered it yet.

In an era where they weren’t able to defeat ‘ubiquitous encryption’ (everything moving to HTTPS) or end-to-end encryption…

… this seems like a convenient way to allow the EU to provide their own CAs who will be willing to cooperate in issuing MitM certs, achieving the goal irrespective.

Certificate Transparency will surely help with this, assuming that they don’t try to override that too, but I’m curious to research now and see if the EFF is taking note of this.

Aside from the details of this specific proposal it seems to me that government mandates on what end-user software can do is uncharted territory. How would this affect open source browsers ? The most widely known browsers (ie. Chrome and Firefox) aren't even traditional commercial products since they are distributed for free and derived from open source code.

Having government mandates on what end-user software can do is not uncharted territory. Some examples:

https://en.wikipedia.org/wiki/Export_of_cryptography_from_th...

https://en.wikipedia.org/wiki/GIF#Unisys_and_LZW_patent_enfo...

https://en.wikipedia.org/wiki/General_Data_Protection_Regula...

Your examples don't support your argument very much.

Export controls on cryptography were eliminated decades ago as it became clear that they made no sense and were counterproductive in a world where the Internet exists.

Patents are a legal concept that has been in existence for centuries but enforcement is only possible through civil court proceedings between private parties. Governments are not directly involved in enforcing patents. Moreover there have been few if any successful patent lawsuits against open source software.

As for the GDPR, from the wikipedia page you linked it appears to apply to enterprises that process data, not software. I don't think the GDPR attempts to restrict what data a software program can process locally on your computer. It only comes into effect if that data gets sent to some company.

> Export controls on cryptography were eliminated decades ago

According to that Wikipedia page, some restrictions remain today. Even if there weren’t anymore, I wouldn’t call it a territory uncharted because we haven’t been there for a few decades.

> Governments are not directly involved in enforcing patents.

Not directly, no, but they decide on what’s patentable and what’s not, and on how long patents hold. For example, AFAIK the periodic Disney patent extension doesn’t come about from civil court proceedings between private parties.

> I don't think the GDPR attempts to restrict what data a software program can process locally on your computer

Similarly, I haven’t read the proposal, but don’t expect this regulation will forbid you to use any kind of certificate locally on your computer, or between your own computers.

> It only comes into effect if that data gets sent to some company.

That, IMO, is limiting “what end user software can do”.

What's the problem they're trying to solve here? Is it just lobbying fortthe return of special EV certificates or is there actually some kind of government issue that these people are trying to solve?

Enabling man-in-the-middle attacks for exfiltration.

You mean something like cloudflare?

I find mitm fears related to CAs rather funny when everybody uses cloudflare anyway.

cloudflare is an entity that you can choose to trust with your data. It's not any different than how your hosting provider/cdn can "MITM" you as well. The issue people are having with this regulation is that the government is forcing you to trust some unrelated party.

You are forced to trust cloudflare too as you can't do business without DDoS protection anymore. I'd prefer such trust to be rooted in my (EU) jurisdiction.

>You are forced to trust cloudflare too as you can't do business without DDoS protection anymore

At best, you're "forced to trust" a DDoS protection vendor, not cloudflare specifically. I'm sure there are DDoS protection vendors that are "rooted in [...] (EU) jurisdiction". Also, switching between such vendors is pretty trivial, much easier than trying to trying to lobby the government into stopping surveillance.

If you can't even name one example without extensive research then it's likely not so trivial. And where you get the firm conviction that both cloudflare and existing CA's aren't tapped already?

>If you can't even name one example without extensive research then it's likely not so trivial.

If I had to name one off the top of my head it would be ovh.

>And where you get the firm conviction that both cloudflare and existing CA's aren't tapped already?

"why don't you let the government mandate encryption backdoors? Your messaging app is probably already backdoored already"

"let's avoid having any laws governing the internet, because all governments are evil, corporations are good and ignorance is freedom!" /s

But my argument against this regulation isn't that "all governments are evil, corporations are good", it's that with respect to certificate authorities, corporations are largely doing a pretty good job, in contrast to governments who are pushing encryption backdoors. If google/mozilla/apple are abusing their position and harming users through their CA programs, then I might consider regulation. But for now it looks like a power grab.

Again, how can you so sure google/mozilla/apple are doing a good job when they are not transparent or accountable to public? We have no idea so we happily live in illusion that they aren't pushing anything nefarious. Whereas the public lawmaking in democracies is messy, fueled by outrage and visible special interests.

>Again, how can you so sure google/mozilla/apple are doing a good job when they are not transparent or accountable to public?

I can be sure they're doing a good job, because I can see them doing a good job, and have no reason to believe they're suppressing any bad news. Are you making a theoretical argument here? In other words, are you simply claiming that the current state of affairs is bad because google/mozilla/apple can go rogue and do bad things?

As for transparency, AFAIK both google and mozilla have their certificate policy discussions available to the public, eg. https://groups.google.com/a/chromium.org/g/blink-dev/c/eUAKw...

>We have no idea so we happily live in illusion that they aren't pushing anything nefarious

This sounds like argument from ignorance. ie. "we have no idea whether they're not pushing anything nefarious, therefore we should assume they're pushing something nefarious". If you think they're acting nefariously, by all means mention it here. Don't go beating around the bushes with vague accusations that they might be doing something bad, and use that to justify the government stepping in.

But vague accusations of government doing something bad are okay?

Yes I am simply claiming that the current state of affairs is bad because google/mozilla/apple can go rogue and do bad things. I'm not claiming they necessarily are. But history teaches us that great power brings corruption, they aren't so much different from Microsoft some decades ago.

> But vague accusations of government doing something bad are okay?

The government's past and current push to ban encryption (or similar efforts) aren't "vague".

>Yes I am simply claiming that the current state of affairs is bad because google/mozilla/apple can go rogue and do bad things. I'm not claiming they necessarily are. But history teaches us that great power brings corruption, they aren't so much different from Microsoft some decades ago.

You have to trust the browser makers regardless. If you want to prevent browser makers from going rogue (eg. sending your browsing data directly to them), you'd need far more comprehensive regulation than this. The only thing this does is force your browser to trust additional set of entities. In that respect, this legislation is quite pointless, because we haven't seen browser makers unfairly exclude upstart CAs. I suppose the regulation might be worth keeping around just in case, but I'm not really convinced of the cost/benefit. As I see it the benefit is that if browser makers decide to go rogue and exclude new CAs, this will allow EU and/or member states to issue their own certificates. The downside is that EU and/or member states can issue TLS certificates to MITM connections. Am I missing anything?

Now I have to trust the EU and Cloudflare.

As much as I despise Cloudflare, at least in Germany the pandemic has shown that authorities can't be trusted with data even if taken with the best of intentions.

Nobody can be trusted with data.

It's cronyism, pure and simple.

Certificate authorities like Camerfirma[1] were revoked from all browsers because of numerous policy violations. This legislation is intended to force browsers to accept their certificates.

[1] https://wiki.mozilla.org/CA/Camerfirma_Issues

I don't really understand how the EU can enforce this. If the major browser vendors threatened to prevent their browsers from being able to be used in the Eurozone over this they would have to back down because without the big four browsers almost every single person in the EU would basically be unable to access the internet. It would cause chaos. All four companies are based in the US as far as I am aware so it's not like the EU could do much about it.

As an English citizen I sure as hell wouldn't want this to happen though.

> If the major browser vendors threatened to

This won't happen, so none of what comes after matters.

> As an English citizen

England is not part of the EU.

> England is not part of the EU.

So? Stuff like cookie consent banners are the EU's doing, and they happen worldwide now.

No, they're not. No need to ask consent if you don't use them (or use only ones you need). The fact that you see them everywhere is a purposeful piece of malice by each and every website you visit.

Nothing like this would happen, the market is to big. And money is all the big vendors care about.

"That was Feisty Duck, reporting on some misguided QWACs that are ruffling feathers, outside the EU headquarters in Brussels. And now, over to you for tomorrow's weather, Dewey -- will it be nice weather for ducks?"

(Burning karma for duck news jokes)

Presumably the browser vendors could require this new EU thing and a certificate from someone they trust.

I wonder what that means for the most important websites on the Internet (those of private individuals hosting their own websites on their own computers on their own Internet connections from their own homes).

just bureaucracy refusing to die as the tech moves on, nothing new I suppose...

I don't want foreign companies to tell me what to trust, i believe in the government I elect, and their representatives in the EU

Yes mozilla, I don't trust you

And btw what you do is called political interference, you should be ashamed mozilla

> is causing concerns

hahaha, what's the goal here? spreading fear and misleading people into thinking that's how they should feel?

typical from mozilla, trying to save their sinking chip

> the certificate authorities that would issue these certificates would be decided by the EU member states, and browsers would be forced to accept those

It's not a question of trusting the EU, but of user autonomy. Even if the EU were 100% trustworthy, that shouldn't let it force users which certificates they trust.

If I phone a friend, but his voice sounds off, and his manner of speech unusual, so that I suspect he's being impersonated - what business does any government have telling me that I must trust it is indeed my friend on the line?

> Thus, if implemented, these plans could mean that browsers would be forced to give special treatment to TLS certificates that have been issued by entities held to a lower standard than the other certificate authorities.

I'm not clear on the "forced" aspect of all this EU regulation. How does the EU force Firefox/Mozilla, Google Chrome, or any software in existence to recognize the EU's list of certificate authorities?

Sounds like compelled speech. How the EU plans to force somebody to write source code to open source projects is an amusing idea... Would I break the law if I forked Firefox, and chose not to include the EU's certificate authority? IF yes, how are they going to compel my speech outside the EU?

Sound like the EU will have to create their own web browser.

I know nobody likes this idea, but I believe the best solution is a curated list of companies, products, and their websites. You provide a browser add-on that displays the list, and let users search through it and click on the company/product they want.

Here's some ways to implement this:

1. Have CAs publish a list of domains and company names. This is problematic because lots of companies have different domains and legal entities, and it's very likely an attacker could list a company that sounds like another company with a malware domain.

2. Volunteers register companies and their domains and brand logos in a Git repo and companies can request changes if need be. They'll be explicitly only trying to list domains that people have asked for and scrutinize them to prevent adding phishing sites.

The latter would have some problems I'm sure, but it gets closest to what users want, which is "show me Bank websites" / "show me Bank of America's website". By having a curated list and some strict policies on managing the list, it will probably be much safer and easier than any other method that exists today to find a company website. Additional techniques could add integrity, like a reporting mechanism, and a star feature to favorite specific sites so they come up first with the number of times they've been starred.

There are several problems with this idea. The first is that of scale - the list of companies along with their associated products and proven-correct official web site(s) would be a nightmare to maintain. I would not trust it to be accurate just because of the scale. One possible solution would be for the company registration with government to include a list of web sites, but then you have the problem of merging company registers from multiple countries, and trusting government. Secondly, allowing volunteers to provide updates to this list allows an attack vector. Thirdly, it is trying to solve a problem that doesn't exist.

It's still best and simplest for a certificate to be tied to a domain, not a company. The last several years has shown that attaching a certificate to a company is ignored by users and doesn't add any security. Attaching a certificate to a domain is something that can be done with a high level of security and assurance, but attaching it to a company (or product) cannot be as securely done, and is information that is less useful.

We can create a toddler net and for some applications that would provide benefits. It could use the same infrastructure as the common net and can leave it completely unaffected. You would just need specialized web clients.

Why would you need specialized clients?

So that those who want toddlernet won’t be exposed to the rest of the Internet.

Their clients will only display todsites.

A browser add-on could do that with no other modification to the 'client' browser (if that's what the user wanted, but that's not what I'm suggesting)

This sounds like you're giving preferential treatment to companies over non-companies. Why would companies get preferential treatment on the internet? I don't think that's a good idea.

Reply on news.ycombinator.com