Anti-Terrorism Hypothetical: Bulk Scanning of Hosted Files? (justsecurity.org)
An anonymous reader writes: The tech community has spoken: we don't want the NSA or any other government agency running bulk surveillance on us, and we don't want tech companies to help them. But Bruce Schneier points out an interesting hypothetical raised by Harvard Law School professor Jonathan Zittrain: "Suppose a laptop were found at the apartment of one of the perpetrators of last year's Paris attacks. It's searched by the authorities pursuant to a warrant, and they find a file on the laptop that's a set of instructions for carrying out the attacks. ... The private document was likely shared among other conspirators, some of whom are still on the run or unknown entirely. Surely Google has the ability to run a search of all Gmail inboxes, outboxes, and message drafts folders, plus Google Drive cloud storage, to see if any of its 900 million users are currently in possession of that exact document.
If Google could be persuaded or ordered to run the search, it could generate a list of only those Google accounts possessing the precise file — and all other Google users would remain undisturbed, except for the briefest of computerized 'touches' on their accounts to see if the file reposed there." Zittrain asks: would you run the search? He then walks us through some of the possible complications to the situation, and the pros and cons of granting permission. His personal conclusion is this: "At least in theory, and with some real trepidation, I'd run the search in that instance, and along with it publicly establish a policy for exactly how clear cut the circumstances have to be (answer: very) for future cases to justify pressing the enter key on a similar search." What would you do?
If Google could be persuaded or ordered to run the search, it could generate a list of only those Google accounts possessing the precise file — and all other Google users would remain undisturbed, except for the briefest of computerized 'touches' on their accounts to see if the file reposed there." Zittrain asks: would you run the search? He then walks us through some of the possible complications to the situation, and the pros and cons of granting permission. His personal conclusion is this: "At least in theory, and with some real trepidation, I'd run the search in that instance, and along with it publicly establish a policy for exactly how clear cut the circumstances have to be (answer: very) for future cases to justify pressing the enter key on a similar search." What would you do?
What about false positives - like if a document has been mass-mailed or put as a part of a story etc.?
I an imagine that we would end up into a situation of "guilty unless proven innocent".
If builders built buildings the way programmers wrote programs, then the first woodpecker would destroy civilization.
Other email providers exist, which ones do we force or ask to scan all their documents?
Do we force companies to scan theirs too?
Get developers to add backdoors scanners to all their software?
This isn't a new problem.
Even though it's hypothetical, it's still dumb.
Once the government has the ability to scan files belonging to hundreds of millions of users for a specific document, it might be easy to broaden that. Searches for similar documents. Searches for a standard set of illegal materials - say known child porn images. Searches for copyrighted materials like movies and audio.
Specifically searching for a specific document with a known like to terrorism doesn't bother me, but the extensions do. I absolutely do not want to give the government the right to search for anything illegal - and I don't see a clear way to enforce the distinction.
The innocent have nothing to fear, but there are few absolutely innocent people
Send or receive a known kiddie porn image through GMail and they will tip the authorities. That hash check can be used for anything the government wants to find people in possession of, just hand them a hash and a NSL.
Live today, because you never know what tomorrow brings
How about searching the account of the one person they've identified to find out which other accounts he had mailed that to?
Then the government can get warrants to search those accounts as well.
As long as they are not in another country or otherwise protected or delete all records after a certain time.
Too complicated for me. We should refer this one to Bennett Haselton.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
They *already* do this, not because they're scanning things, but because they index files on their hashes in the first place.
Remember "Dropship"? If you knew the hash of a file on Dropbox, you could "teleport" it into your Dropbox storage by using the API to tell Dropbox that you had a file with that hash locally. Since it got used for file-sharing, that was shut down - but it reveals that for de-duping purposes, Dropbox hashes all your files anyway.
It's a trivial matter to take that file, hash it, do a lookup in the table of files that belong to each user, and produce a list of the accounts. It's probably already been done for copyright-infringement suits.
No bulk scanning required. Just a lookup.
Lets deal with threats like ISIS at their source rather than playing wack-a-mole with our liberties here at home.
The problem with this search is that government is too big and too unaccountable to be allowed that capability. Governments and law enforcement agencies routinely act unjustly. They use violence and threats needlessly, acting as bullies rather than public servants. And they are almost never punished when they commit crimes.
If governments showed humility and served the public, maybe you'd consider letting them search something occasionally. But that sort of government seems like an impossible fantasy these days. So no. Not until they prove they can be trusted -- which unfortunately means probably never.
In the U.S. (where Paris is not, I realize - but neither is Harvard) we ostensibly are innocent until proven guilty. No, no, quit laughing... I'm trying to make a point. The searcher has no foreknowledge that I might be guilty, so they shouldn't be able to look through my "stuff" for evidence of guilt.
I don't see how this is materially different from, say, having permanent access to my home surveillance camera footage and routinely using bots to review them for the image of a rocket launcher. If you have no reason to suspect me of involvement with a crime, you can't just randomly search my house, stop me on the street to frisk me, or search my personal papers.
#DeleteChrome
I was just about to inform my cells where the nuke is located!
As was pointed out by a commenter earlier when Bruce Schneier posted this.
This whole hypothetical is moot and has already been attempted for DMCA and Child Porn cases. This is because Deduplication is a feature of any large file sharing entity gmail included as drive space is not free.
Because of deduplication there will only ever be one copy of the relevant file clusters in existence and a table of assignments for which messages and or accounts to apply it too. Thus given an example of the file or the list of cluster hashes and a simple court order a company can expunge the one copy and/or return the list of holders with their association / upload / download dates.
Now one key issue would be that even a single bit changed in the file (mentioned in the article) would change the file hash and probably 50% of the bits in the specific cluster would flip. But for larger files >10MB it may be sufficient to match a percentage of cluster hashes and then inspect the misses further.
That said a savvy antagonist would recognise the above and suggest ways to defeat deduplication, even without using anything fancy. For a text file, simply running it through a compression algorithm would change it sufficiently and if you use one that does encryption correctly then each encipherment, even with the same key, would result in a different file. Plus since you are not actually interested in securing the file you could include the password as the filename.
And why I dislike framing the argument against this sort of thing as a right to privacy like the EU does. I tried bringing this up way back in the 1990s when rumors of Carnivore began circulating. If you frame this in terms of privacy, then this type of surveillance becomes legal. It's not a person searching your files (or sniffing network traffic) for pattern matches, it's a machine. Only matches are turned over, and the data of innocents remain private despite having been searched.
It's not privacy which is sacrosanct here; it's the right to be free from government searches without sufficient suspicion. The way this needs to be framed is in terms of the limits of government power. If the government has this type of surveillance power, you may joke about the word "bomb" in your post triggering the FBI's monitoring software, but the chilling effect it has on free speech is the same whether the search was carried out by software or by an authoritarian government trying to control the populace.
To put it another way if we've decided that individual freedom is more important than government control, then this is one of the tools we simply cannot allow government to have. It is incompatible with the notion of government for the people, by the people, of the people. If that leads to the downfall of democracy, then so be it. The sole reason for democracy's existence is as a bulwark against authoritarian government control. If democracy self-implodes in this fashion - because people are too scared of terrorism they democratically choose to give government that authoritative control - then we'll either just have to accept that democracy is conceptually a failure, or we'll have to come up with a new idea for a system of government which respects and protects individual liberty.
The problem is that this is the precise definition of slippery slope. As attractive as it would be to scan for such content legally, this is not the kind of toys we want the government to have. Would the government as with a foreign enemy, we should be discussing capabilities, not intentions. The one inescapable truth is that any capabilities of a bureaucratic entity are going to be abused. If you don't want the abuse, don't give them these capabilities took begin with.
This didn't make the DHS smarter. It only made Bruce dumber.
Let's start with his example: the Paris attacks. The Paris attackers plotted everything using... wait for it... SMS. Just about the least-secure communications system ever devised. About the only way they could have fucked up worse would be if they planned the attacks inside a police station, talking to each other with bullhorns. That's not surprising, of course; the criminal geniuses whose masterplan was "get guns and shoot people with them" aren't going to think of using encryption, decentralized communication, or anything else that even the average slashtard knows how to do.
Now let's move on to Bruce's example. So the police capture or kill a suspect, find his place of residence, find his laptop, his laptop is unencrypted, the terrorist masterplan is just sitting there in plaintext, and... that's it? There aren't any other or better investigative leads? Their best and fastest strategy is to ask Google or whoever to scan all the data of 900+ million users? There's no other evidence on the laptop, no "electronic paper trail" from his online communications, nothing useful in his apartment, they couldn't recover his phone, they can't track the gun he used, they've got *nothing* except a mass surveillance dragnet? The cops just gotta twiddle their thumbs for several hours while Google/Apple/Microsoft/Yahoo/whoever process their request and get back to them? The same terrorist who was so smart he covered all of his tracks was also so dumb he left this vital, identifying, incriminating piece of evidence just waiting for the cops to find it?
It took me as long to read about this idea as it did for me to invent a countermeasure to it. Take some JPEG of a stupid meme, append the terrorist masterplan to the end of the file (or just stick it somewhere in the EXIF data), attach it to an email with the subject line "ch34p V14Gr4!!!!," and use a compromised webserver to bulkmail it your co-conspirators (and a few hundred thousand other people). I'm pretty sure even the dumbest terrorist can manage to download a JPEG, open it with Notepad, and scroll past the gibberish until he finds something he can actually read, and meanwhile the counterterrorism geniuses are working their way through a pool of suspects big enough to populate San Francisco.
This is fucking stupid, Bruce. You're asking me to buy some hypothetical scenario where the perpetrators are so dumb that this strategy would work and yet so smart that this is the best strategy that would work.
The Japanese attack on Pearl Harbour wasn't unusual; the US Navy imagined exactly that scenario. Similarly, bin Laden didn't predict the damage from flying a plane into a building; recently published novels did that. Also, the US DHS spent a few years discussing every kooky attack vector there was. What exactly qualifies as "instructions" in this description?
It gets worse: The FBI profile for a terrorist includes possessing a Casio digital watch, or a pocket reference to the US constitution. The slippery slope here, is the government can use any criteria to scream "look, terrorist". Everyone forgets the US government has a lot of difficulty dealing with slippery slopes, instead choosing an all-or-nothing policy.
These are weasel words like "suspicious people" justifying mass surveillance. How many times has a government stopped every car on a highway and searched it? How many times have the police done a house-to-house search of a neighbourhood? Yet, when that personal 'space' is stored on the hard drive of some corporation, ransacking the 'neighbourhood' is encouraged, which sets a precedent: The corporation is responsible for national security; the hard drive can be searched at any time the government has a problem; plus, the government can push the cost of such ransacking onto the corporation.
First and foremost: You think the 'bad guys' haven't already thought of something like this? You really think they're sending out their most secret plans out in the clear, or even sending them out at all? Anyone with half a brain would either encrypt them somehow (either digitally or by more traditional methods), or use stegonography, or hand-carry them, or commit them to memory, not leave a trail of breadcrumbs that any armchair detective could follow.
Second: This would set the precedent to bring about the absolute and total end of even a pretense of privacy for everyone. It would become leveraged for seaches of anything and everything; everyone's lives would in essence be laid bare for any government agency with a half-assed reason for a search. Not much longer after that the private sector would find a way in, and I wouldn't at all be surprised if not long after that, it would be used outright for marketing datamining.
This is a dangerous, stupid idea, and no way in Hell should it ever be allowed to even be so much as discussed as actual legislation.
Are YOU using the TOOL, or is the TOOL using YOU? Think about it!
At least not in the practice of many Muslims. Living near a mosque, the range of clothing on the men coming to it ranges from standard Western to obviously ethnic. If they aren't dressing by that standard for attending the mosque, it's clear that they won't at other times.
Hillary and wall street bankers in jail would be a start.
To be clear on this ... while you may trust President A not to abuse this, that means that you must also trust Presidents B, C, D, etc. Eventually there will be someone elected that you really do not agree with.
And that person will have all the authority you supported for the people you did agree with.
And none of the inhibitions on abusing that authority.
Supposedly the USSR had copy machines etched so that it was possible to track down the source of aberrant materials. A means of tracking is also done with consumer copiers in the name of reducing fraud, but there is no law restricting it solely to that use.The Federalist Papers would be an anathema today.
Exactly how much further down this rabbit hole do we want to go? Yes, it is fine and good that these measures will only be used with the best of intentions, but if the difference between a police state and your liberal democracy is intentions, you are already fucked.
and write about it on my lap top surely because Google can do what they can do, they should let the gov/police access to all Google services that are used by my family, friends, co workers and neighbors? Crack pot much?
Terrorism is a crime just like any other and authorities should only have enough power to investigate it like any other crime.
by TheSpoom (715771) Uncaring Linux user here. I have nothing to add to this but please continue. *munches popcorn*