Hacker News Reader

Comment by jrapdx3 | Hacker News Reader

The article describes the Justice Dept. doing exactly what has been so widely predicted on HN and elsewhere. Doesn't seem like good timing on the government's part to publicly announce the intention to seek numerous iPhone "unlockings". Rather it plays into Apple's argument about unleashing a torrent of court-ordered demands which will have adverse consequences for security.

Legal minds should weigh in, but I'm thinking the only effective remedy is going to be congressional action, to pass law that defines the limits of court discretion re: forcing firms' assistance breaching their carefully constructed security/privacy systems.

This news prompts me to write my representative and senators and strongly urge them to support enacting this form of protective legislation. If there's enough of an outcry from the electorate there's a far greater chance of putting sensible policies in place. The fact it's an election year can only make voicing our concerns all the more effective.

View on HN

It makes total sense that the givernment would use it's forensic capabilities whenever the need arises. The question at hand is: "can the All Writs Act compel a company to proactively defeat its own security measures?" The answer will apply to all cases.

The more important question is "Should the government be able to access citizen's digital data with a court order? And if so, how can that be enabled without compromising the general security of the device?

I suspect we'll ultimately wind up with mainstream device manufacturers maintaining some kind of per-device master key that they turn over when ordered.

This will enable government to access data in the typical case. Including foreign governments. Sophisticated bad guys will continue to use zero knowledge software, which government will attack by other means. The per-device master keys will be compromised eventually, which will force enterprises to upgrade their fleeta but normals won't care.

> Should the government be able to access citizen's digital data with a court order? And if so, how can that be enabled without compromising the general security of the device?

No. And that's both impossible and a massive compromise.

This case helps us tackle that first question. Here the murderer's personal phone and computer hard drives were destroyed — rendering them "above/beyond the law". Just because some data is digital doesn't place it in some special legal realm more important than shreddable/burnable paper or the air secret conversations were spoken into. There are fundamental limits to recoverability. If technology companies are to be forced to maintain vulnerabilities because governments see all their customers as potential terrorists, the industry is doomed.

The real problem here is although terrorism will never touch the average citizen anywhere near the extent of other tragedies like illness, accidents or natural disasters, the media treat it like it's the single most important issue — making people fear for their lives is good business. I'd die before sacrificing freedom of speech every time, but the news business just seems too like racketeering. We need to fight the fear.

> terrorism will never touch the average citizen anywhere near the extent of other tragedies like illness, accidents or natural disasters

This argument ascribes zero weight to the injustice of terrorist attacks. Your logic--that a death is a death--does not admit distinguishing between someone dying in a freak accident, someone being killed by a drunk driver, and someone being murdered in cold blood. It's all the same.

You're ignoring a very fundamental aspect of human psychology: people view a death very differently based on the intent of those doing the killing. Unlike murder, terrorism isn't just an attack on one person. It's an attack on the values, religion, economy, and lifestyle of a whole society. That's why people weigh it so heavily.

Possibly. An individual murder means it is unlikely to have an impact to you. A serial killer will often impact a demographic in a city. A mass murderer will often impact a demographic in a large area or country. A terrorist will usually target a demographic in many countries, but on a much smaller scale than a mass murderer.

I currently think a mass murderer would be a greater threat to the world collectively, but the terrorist triggers the fear that any location might be attacked. Hence, while much less destructive, has the "me" factor that pulls heart strings of society at large.

Why not? Police can search your house or seize your mail with a court order. Why can't they search your phone or computer?

They can search my phone and computer with a court order. What they should not be able to do is force companies to compromise proper encryption so they always have the ability to find something.

That's not the only angle here. Apple was asked to aide the FBI in attacking a phone, not to design bad crypto. (They may have also been asked to design bad crypto, but that's not what is happening here)

Except that the authorities do want companies to use bad crypto. The only reason they've had to fall back on demanding an attack vector is because they haven't yet been able to force their preferred solution (bad crypto) to be implemented.

Demanding an attack vector should be seen as the same concept as demanding bad crypto, because the intent behind the request is the same. They're trying to convince us that these are different requests, but the end result is the same. A workaround to attack good security is the same as having bad security to begin with. I can't imagine why anybody would think that "bad crypto" and "attack vector" are not very nearly the same thing.

But what you are forgetting is that Apple has been fully compliant and cooperative throughout this investigation. The problem with building a backdoor into a highly encrypted security system gives pathways for others to find the same backdoor. If other hackers knew there is a for-sure way to gain access and hack an iPhone, they will find that path. With today's plethora of technology, a line needs to be crossed in order to protect our privacy. We hold so many personal details inside of our phones and if by some chance the backdoor were to be released, chaos and panic would run ramped. I can understand completely why Apple deems this process "too dangerous".

The pathway is obvious- build a signed image that lets you guess unlimited passwords at maximum speed. Apple doesn't have to do it to make it apparent it would work. The avenue is already in use:

As many jailbreakers are familiar, firmware can be loaded via Device Firmware Upgrade (DFU) Mode. Once an iPhone enters DFU mode, it will accept a new firmware image over a USB cable.

The special "backdoor" Apple has access to:

Before any firmware image is loaded by an iPhone, the device first checks whether the firmware has a valid signature from Apple. This signature check is why the FBI cannot load new software onto an iPhone on their own — the FBI does not have the secret keys that Apple uses to sign firmware.

http://blog.trailofbits.com/2016/02/17/apple-can-comply-with...

As for "the attack firmware could leak"- well, so could the signing keys.

Because they can't. A court order can't let them fly. Nor can it compel you to build them wings. (Though you may have to buy them a can of Red Bull.)

The police were already searching you and your house so we enacted rules to try to control that. Those rules didn't enable the searching - they placed restrictions on the applicability of evidence to reduce the desire to search improperly.

If they can get in. Suppose your house is made of some material (10' thick steel?) which cannot be broken into without a use of a nuke (bear with me....). Is the government allowed to nuke that neighborhood just to get into the house?

>I suspect we'll ultimately wind up with mainstream device manufacturers maintaining some kind of per-device master key that they turn over when ordered.

It will be impossible to secure that database. Any number and size of bribes would be worth spreading around to gain a dump of it. Once I have a dump, you're up shit creek because you're faced with changing every key of every device you've ever sold to resecure them. However, any key changing mechanism you create is an exploitation vector criminals can use to randomly scramble their "master key" and not report the new one to you, and again, any certs you use to secure this process will become v cost effective to bribe for. So we're left with a situation where one breach screws everyone and it hurts legitimate users way more than the criminals it is targeted at.

Back doors are back doors man. Anyone can use the door handle, once it exists.

These kind of master keys already exists so it is not something hypothetical. I am talking about the private keys of trusted root certification authorities. If it would be as easy to get them as you say nobody would probably use TLS since it would be useless.

Well, TLS with "trusted" CAs is useless, for extreme threat models.

Care to elaborate?

You're vulnerable to MitM if adversaries have issued fraudulent certificates for sites that you use. And we know that this has happened, more than once.

And that'll only get worse if the FBI succeeds in compelling a US technology company to sign something they don't trust.

http://thenextweb.com/insider/2015/04/02/google-to-drop-chin...

it's not tinfoil to note that there are CAs under the control of authoritarian governments.

I think the point is that it's easier to spoof a trusted cert than to crack the encryption

> The more important question is "Should the government be able to access citizen's digital data with a court order? And if so, how can that be enabled without compromising the general security of the device?

The government already does have access to these citizens' data. That data simply has the quality of having been encrypted.

If I have printed encrypted text and lock it in my home safe, when the government gets a court order to search my house, they'll have complete access to those documents. They can't compel me to decipher those documents though, it's on them to break the encryption.

It's the same situation here. Here, Apple may be able to perform that decription service for the government.

The question to me is, should the government be legally able to press a person or firm into service?

I think a better analogy for this sort of thing is an unknown location. It's like you took some unencrypted text and hid it in a tree in a national forest. Everyone strongly suspects that the unencrypted text existed in the past. It's just that only one person knows how to access that text in the present.

The problem with the locked box analogy is that it implies that the cypertext is in some way equivalent to the plain text. In practice that isn't really true. If you don't have the key you don't actually have the plain text. Nothing says you can't look for it if you want. You could even hire a company to help you look. Can you force a company to look if that search would be against their business interests?

Ah, I'm not using the safe in the above as analogy to the cyphertext; in both cases I'm talking about encrypted text being stored (printed physical form or digital data form).

You scenario is wonderfully worded and poses a great dilemma. I personally believe that the government should not be able to press a person or firm into service. In today's world of technology, we tend to forget that what we have on our phones is simply an electronic copy of paperback versions. "Printing out" our phones in the encrypted way you refer would not be any use to anyone but the primary owner of the data. Just because we have the means to decipher documents does not mean we should and it should be ultimately up to the person/firm to decipher said documents.

If the encrypted data is stored on a hard drive in that safe, the government can compel production in some cases, blocked mainly by a legal quirk involving the Fifth Amendment:

https://en.m.wikipedia.org/wiki/Key_disclosure_law#United_St...

IANAL, but I think your original hypothetical would be roughly analogous. I think the difference in this Apple case is essentially the amount of effort required (producing a password versus actually writing [a modicum of] new OS software), but I could be wrong.

My NAL understanding there was that, in cases where the government has already seen the evidence, decription is not self-incrimination. But in general excepting that, they can't compel the owner in the possibility that doing so would force the owner to self-incriminate.

The other question is if it's legal for a company to design their devices so that it is technically impossible for them to comply.

Is there any practical reason why the secure enclave could not be programmed such that any firmware update to the enclave requires the PIN, and if not it erases the key. That way Apple is still able to make updates to the enclave, but cannot be compelled to bypass security restrictions on locked devices.

Pure speculation: I wonder if v2 of the Secure Enclave was going to do just that, but Apple wanted to make sure there weren't glaring software flaws in v1 before they lock themselves out of the chip for good.

It would be difficult to design a law in a Napoleonic system that disallowed that(designing their devices so that it is technically impossible for them to comply), although in the American system I guess it would be possible.

We've come close to this already (mandated Clipper chip). Which would have been a disaster, in many areas.

IIRC, the ITAR rules even prohibited systems that lacked crypto, but that had pluggable APIs which permitted crypto to be added later (e.g., by the end-user). It's hard to see how legislation to the same effect would pass constitutional muster, but it's a gamble, and the administration has little to lose.

>Should the government be able to access citizen's digital data with a court order?

I don't think that's a good question at all. Regardless of the ruling, they aren't going to up and declare encryption illegal. You can't throw a legal requirement at math to stop working.

This is dangerously naive. You think the government can't do something illogical? You just wait for it. You think the government can't outlaw the use of something, even if it's an ineffectual half-measure that does more harm than good? Were you alive during the "War on Drugs"?

If you studied history they passed prohibition that banned alcohol. It did not work as bootleggers and gangsters would make their own alcohol and sell it at hidden clubs. Stores would sell fruit juice next to sugar and yeast and all the customer had to do was find a way to mix them to make alcohol.

They eventually had to repeal Prohibition.

When they banned drugs, they did the same thing. It is prohibition all over again. Criminals and gangs got into dealing drugs for money and people can't give them up. The war on drugs doesn't work because the CIA makes money from drug lords, and gangs and criminals make a lot of money selling drugs on the street. I managed to grow up without getting into drug but during the 1970s and 1980s there was a lot of peer pressure into doing drugs with others when I was growing up. It has become a social thing.

Many prisoners are in prison for using drugs. So our prison system is over crowded as a result. All the war on drugs did was imprison drug users and not go after the gangs and dealers.

The government doesn't get privacy rights. So they violate privacy in order to crack encryption on an iPhone via a backdoor or master key or modified iOS that gets rid of passcode penalties and brute forces the passcode. They only see getting the data off the iPhone to use as evidence, and don't care that it causes iPhone users to lose their privacy and make iPhones less secure.

The government isn't logical about these things, judges and politicians are for the most part not tech savvy enough to understand the technology and the need for privacy. They studied law, and other things, and not technology in the most case to understand how it works. They see getting evidence on terrorists as a priority and don't care what rights it violates. They can pass a law that pressed criminal charges at Apple employees and management for not helping out and try to force them to do what they say. We faced the same issues on the war on terror that required domestic spying and collection of metadata via the US Patriot Act.

Maybe if normals care enough about privacy to make a big deal out of any attempt to breach privacy it would be different.

but as it is they just shrug and say "i got nothing to hide"

smh

Not just "I got nothing to hide", but "You shouldn't be able to hide anything", which is scarier.

Why "were"? We are still alive, humanity is still suffering from the "war on drugs".

> You can't throw a legal requirement at math to stop working.

Well, it's not stopped them trying in the past [1] - and we now live in a world where hilariously/terrifyingly certain numbers are just flat out illegal [2]

In my mind, a large part of the problem comes from the fact that the people making these rules not only have no understanding of the underlying science, but they aren't even expected to _try_ and understand it, even at a cursory level.

Which in itself might not actually be so bad, except we've seen examples time and time again (especially in the UK) where ministers feelings or beliefs trump expert opinion and/or hard science [3]

[1] Indiana Pi Bill (an attempt to legislate math): https://en.wikipedia.org/wiki/Indiana_Pi_Bill

[2] Illegal Numbers, see also Illegal Primes: https://en.wikipedia.org/wiki/Illegal_number

[3] Medical science vs war on drugs, see David Nutt: https://en.wikipedia.org/wiki/David_Nutt

[3a] Current "Snoopers Charter" discussions where industry experts still don't understand what's actually being legislated.

Not only that, seeing "the government" as a single entity is at best naive.

Can a schoolteacher access a citizen's digital data with a court order? Can the FBI access a citizen's digital data with a court order? Can a court access a citizen's digital data with a court order?

Those questions are fundamentally different and have varying support in lawmaking.

> It makes total sense that the givernment would use it's forensic capabilities whenever the need arises.

Assuming you mean "its" then apple's capabilities are not the government's. The government could physically attack the chips holding the data instead.

They're just risk-/effort-averse and would like others do the work for them.

We also will wind up (already have) with more smart terrorists who hide data in better places. So the only people who lost are we (again).

As a devil's advocate, couldn't we say the same thing about guns?

We can. But we'd still reduce the accidental gun deaths tremendously—as well as killings by people who weren't previously criminals. Yes, we probably won't get the guns away from criminals—there are too many guns, and they've been around for too long. But we'll still reduce the deaths.

> "Should the government be able to access citizen's digital data with a court order? And if so, how can that be enabled without compromising the general security of the device?

IMO yes and here is how I'd implement this given a secure enclave:

1) Have Apple generate a device specific key that unlocks the device without pass code.

2) The key gets encrypted with an Apple held master key and printed on the inside of the device, e.g. a sticker on the chipset. Note that at this point, the device is as secure as before if we assume that someone who gets the Apple certifier key can exchange the secure enclave software at will.

3) Apple should ensure that the unencrypted key gets deleted from all of their systems after it was printed, to ensure plausible deniability.

4) The process for accessing the data on a specific device becomes simple - law enforcement needs to be in possession of (A) the device and (B) a court order. Apple will decrypt the key if forced by court order.

I think, we can assume that up till now, Apple has its private keys protected on some disconnected machine in a HSM, perhaps multiple HSMs, that are only accessible with very restricted access. When there is a gold iOS build, some of the key personal will go through a security procedure to get the build signed.

Now suppose that such a procedure is put in place. And you get requests from US courts and Western European courts[1]. Once it's there, there will probably be hundreds, perhaps thousands of daily requests. It suddenly becomes undoable to keep the same security procedures. More people need access to the master key to handle all the requests for which there is a court order. Or even worse: there has to be an semi-automated procedure to handle such requests.

In any case, any such procedure will weaken the security of iOS devices significantly. Since Apple's master key is so extremely valuable, it will be an extremely attractive target for attacks (in the semi-automatic scenario); or it's easier to find a corrupt person to leak the keys.

Since people store anything from personal photos, access to bank accounts, to business plans on their phones, we should move into the direction of making crypto systems stronger, not intentionally weakening them.

[1] Let's avoid the slippery slope that this would trigger for not-so-nice regimes for a moment. Although, once such a backdoor is there, it will happen.

I concede, these are good points. I'd like to add though, that it would be better overall to have a solution that at least tries to limit access to specific devices, rather than Apple being forced by the government to build in general backdoors. But I can't really think of a better solution. Maybe have a hierarchy of keys, one per some serial number suffix, each in different servers in different buildings and accessed by different people, so if one gets lost it at least only affects a fraction of devices on the market?

I think that we as a society (though I am not American) should decide first whether we think backdoors for court orders are worth the risk at all.

Let's be honest with ourselves, the chance to be killed by terrorists is minuscule compared to other factors that are under human control (air pollution, traffic incidents, homicides).

If you are looking at non-terrorist criminal activity, one has to wonder if you don't have more serious problems than unlocking a bunch of iPhones when your homicide rates[1] or traffic-related deaths[2] are higher than nearly any other Western country.

Terrorism and crime are just easy distractions to get more power.

[1] https://en.wikipedia.org/wiki/List_of_countries_by_intention... [2] https://en.wikipedia.org/wiki/List_of_countries_by_traffic-r...

HSM - Hardware Security Module?

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

And if the Apple master key is ever revealed or stolen, everyone's phone can be decrypted. So yeah, this plan still compromises the general security of the device.

As is the case with Apple's software certifier key. AFAIK this is what we currently have to assume since Apple doesn't specify how the update process for the secure enclave works. If this is the case it doesn't change anything, the certifier key is already like a master key.

Replacing the firmware (Secure Enclave or not) with a version that doesn't include rate-limits and auto-wipe isn't the same as having a crypto backdoor. Right now, your data is still protected by the complexity of your passphrase. This is not true for your scenario.

No. Whatever backdoor you enable for the US government can be used by other governments. When traveling internationally I would rather just deal with the minuscule increase in terrorism risk in exchange for a phone that can't be decrypted by anyone but me.

> "can the All Writs Act compel a company to proactively defeat its own security measures?"

I think there's a qualifier on there that you're missing. It's not just if the All Writs Act can compel a company to undermine their security measures. It's "if a company can undermine their security measures, can the All Writs Act compel them to do so?"

I think the answer is, arguably, "yes."

But that's a meaningful distinction because the 5C lacks trusted execution, and newer iPhones have it. I just read an article where Apple claims the precedent set in this case could force them to undermine phones with trusted execution. And that's probably true, too, IFF they can undermine it themselves.

But the All Writs Act certainly cannot bar a company from creating something that cannot be undermined.

So they'll make firmware that can't be modified without wiping the device.

I wonder if there's a way of using software licensing to bypass the issue. Consider this. What if there was a piece of encryption software where the copyright was held in the UK (i.e. not in the US). The key thing about this encryption library: it has a special license. One of its clauses says that it may not be edited, it may only be compiled by clang, may only be distributed in the form that results from a good-faith compilation under clang, and that - as a condition of use - the distributor may not implement workarounds. The US court would have no jurisdiction for ruling on the actions of the copyright owner. And Apple could defend its behaviour by saying that intellectual property law means that it is not within its rights/mandate to modify the included software.

A US court can order Apple to do what it likes inside US jurisdiction, including things that would normally be illegal in the US.

Citation? It seems pretty unbelievable that a US court can order someone to do something that violates US law.

What is it you are seeking a 'citation' for? This is so obvious nobody with a legal background would question it, let alone write about it. The gp is so silly it doesn't warrant discussion. Law is not a logic game with a closed rule set, it's nonsensical to reason based on that assumption.

Just asking for an example. I don't have a legal background. No need to get defensive about it.

The government can overrule private contracts, and does so on the daily.

The best way (IMO) is to make it unbreakable, and off by default. Then the people who care about privacy can have it, and the justice department can be happy most of the time.

But does that imply that the technically inept don't deserve privacy, even if they want it? I think it does, and is therefore an unviable solution.

But I am sure a large portion of the people involved in activities where the government wants their info will have it turned on. I'm sure if Snowden's phone was encrypted they wouldn't go "Oh well ~90% of americans dont have encryption turned on so we'll just be happy with that."

Snowden is a poor example here. He worked in intelligence, a field the government would logically and rightly expect to value encryption. This, Snowden would fit in by using encryption here. He'd stand out if he didn't.

Then the people who care about privacy become targets.

They are already targets.

If it's opt-in, then encryption on your phone leads to 'what are you trying to hide?'

If it's on-by-default, then everyone has their phone encrypted, and it's never suspicious.

Everyone cares about privacy. What are you saying exactly?

Perhaps the middle part of the argument is missing? People who care about privacy == people who use encryption. People who use encryption already have all of their communications flagged and stored, if the rumors are true.

I disagree. This surely applies to the tech sector, but what about the wider world? Many people have all sorts of privacy habits (close curtains, have tall fences and hedges, they shred their mail or don't discuss wages, etc) yet don't understand or practice anything similar in the digital world.

Why do you think that this is the best way? Or do you mean the best way for the Justice Department?

It's not actually his phone. It was his "work phone" and it belongs to the San Bernardino County Department of Public Health, who is very willing for the FBI to read the data.

Howabout the question is, should Congress repeal the All Writs Act.

Probably not. But they (or the courts) should make it clear what actions it can be used to compel, and what actions it can't be used to compel.

What's alarming is that sites like The Verge are posting articles like this [1] essentially condoning the FBI's position.

I would have expected them to know better.

[1] http://www.theverge.com/2016/2/23/11100152/apple-fbi-terrori...

Conspiracy theory: The Verge is still angry that Apple enabled content blocking, thus allowing users to remove most of their adtech bloat.

I read the entire article and I cannot find basis to support your position on it. Was it edited?

> "Rather it plays into Apple's argument about unleashing a torrent of court-ordered demands which will have adverse consequences for security."

I wonder if that would have negative economic consequences for the US, e.g. tech companies that value privacy moving to Europe or elsewhere and customers avoiding US tech products.

If so, i imagine that would probably be the best reason for the US government to not proceed with this?

> Doesn't seem like good timing on the government's part to publicly announce the intention to seek numerous iPhone "unlockings"

That was my immediate gut reaction, that it looks bad because Apple objections are proving true.

But now I wonder if the timing maybe works to their advantage, and the admission they want to use phone cracking everywhere is perhaps a very smart strategy. To technologists and privacy advocates, this might look like an admission of guilt. To the public at large, and the entire country, this might actually be a mountain of evidence that phone cracking is absolutely necessary, something we can't do without if otherwise phones are going to become unreadable by the "good guys".

For better or worse (and IMO it's worse) this might be a very strong argument with very good timing.

I disagree. I think the only effective remedy is strong, successful cryptography.

the only effective remedy is going to be congressional action

We are well, well beyond that point. This will never happen, given the authoritarian and fearful bent of the electorate and the elected.

The 3% solution is the only truly effective remedy.

The best respons is already a reality: iPhone 6 with Secure Enclave - or making it impossible to unlock the phone with a hack and with brute force.

I hope Google and Microsoft follow suit, and make this a requirement for future versions of Android and Windows Phone.

Nope, Secure Enclave is updateable, even Apple said so.

Do we know what can the software running on the secure enclave do exactly? If the software itself doesn't have access to the master key (a.i. can't read it directly), and if the slow authentication is inherent from running the algorithm on the chip, rather than a having been slowed down artificially, then even an update to the secure enclave would not compromise the security of the system.

Do we know exactly what the secure enclave software can do and what it can't do?

> if the slow authentication is inherent from running the algorithm on the chip, rather than a having been slowed down artificially

There is no known cryptographic algorithm that provides inherently increasing times. The slowdown is entirely artificial, and it is believed that an update could remove it.

I was under the impression that the algorithm (running on the secure enclave?) takes at least 200ms. Yes, obviously longer times are artificial, but my question is whether the 200ms baseline is artificial or not.

Of course this doesn't really matter at all if modified software could just read the key from the chip. That is the most interesting question. So, can it? Or can the software only provide data to the chip, to do the algorithm in hardware, but the software can't read the key burned (?) into the chip.

Their hash algorithm takes ~80ms per attempt. All crypto happens on a dedicated AES engine. Files are encrypted with a key derived from UID (device key which cannot be extracted via firmware) and passphrase.

So if the key can't be extracted via the firmware, someone updating the firmware, even with malicious intent, should not be a problem, am I correct?

With a sufficiently complex passphrase, it's not a problem. With a typical 6-digit PIN, disabling the artificial delay and auto-wipe (which is possible via firmware updates) is all you need for a successful brute-force attack.

Thanks for the clarifications.

The 200ms hash baseline (on all up-to-date iPhones) is not artificial, but the increasing time is.

Unfortunately, I don't believe the answers to your other questions are known to anybody other than Apple right now.

It does not matter for the case at hand, which is FBI wanting to bruteforce the code without 10-try erase.

Yes, it does not matter for the case at hand, but it is very important in general, and it has been brought into attention up the thread.

Reply on news.ycombinator.com