News for hackers. Stuff that matters.
Re:Orin Kerr: Why I Am Representing Auernheimer Pro Bono on ... by AnthonyMouse on Friday March 22, 2013 @02:13AM (#5421623)
Here's the problem with all of this: If you send a sufficiently vulnerable server a specially crafted request, you can get it to come back with "200 OK" and a list of everybody's credit card numbers. Conversely, if you're an employee of Foo, Inc. and you sign in to a secure server with your personal account and try to do something privileged, it's going to come back with "401 Unauthorized" and not give you anything even though you are actually authorized, and if you then sign in with your employee account it will allow you to do that thing.

As a general rule what the machine says has a very strong relationship with whether or not you're authorized to do something. The issue is that if you're not authorized then a properly functioning machine just won't let you do it, which means that it seems impossible for anyone to violate this law against a server that is working properly.

The only way anyone can be capable of breaking this law is if the server is not working properly and allows them to gain access without authorization. Which is why "unauthorized access" is such a vague and hopeless disaster. Going by the normal mechanism for determining whether you have authorized access, namely whether the server allows you to do something, would mean that no one could ever commit the crime, because either you never actually gain unauthorized access since you're prevented by a server with sufficient security, or you succeed in convincing the server to let you do something which under normal circumstances implies that you're authorized.

Seemingly the only way anyone would ever be convicted is based on a pile of circumstantial nonsense about how the defendant should have known they weren't authorized to do something that the server allowed them to do, even though normally you are authorized to do anything the server allows you to do.

So it becomes a de facto law against "doing bad things with a computer" -- not a specific prohibition against anything in particular, just something you stick to anybody who you don't like, because hey, if you did something "bad" then it wouldn't be authorized, right?

Re:Orin Kerr: Why I Am Representing Auernheimer Pro Bono on ... by flyinRyan on Friday March 22, 2013 @07:37AM (#5422336)
> If you send a sufficiently vulnerable server a specially crafted request, you can get it to come back with "200 OK" and a list of everybody's credit card numbers.

If, if, if, if. What Weev did was spoof what kind of client he was using. That's it. What you're suger coating here is using exploits to break into a secure system. That is, you encounter a secured system and find a way to circumvent that security. For the data Weev encountered was there any possible way to get a 401 response for the URLs?

Re:Orin Kerr: Why I Am Representing Auernheimer Pro Bono on ... by AnthonyMouse on Friday March 22, 2013 @10:11AM (#5422967)
>What you're suger coating here is using exploits to break into a secure system.

No. Because using "exploits" (not a legally defined term AFAIK) doesn't necessarily mean that access was unauthorized. If you're the sysadmin for a remote server that you suddenly discover you can't login to with your account and that people are complaining that it's sending spam, and you smash the stack on a vulnerable application running on the server in order to regain control and shut it down, I should hope that wouldn't be "unauthorized access" and subject to criminal penalties.

And then there's the fact that "exploit" is a fuzzy and undefined thing. Is changing "userid=4833" to "userid=4834" to get another user's account not an "exploit" but changing "userid=4833" to "userid=0" to get root access is? What if the maximum userid is 65535 and if you use "userid=65536" then it rolls back around and gives you root because it's equivalent to "userid=0" but doesn't get rejected like "userid=0" would? This is no way for a criminal law to operate.

>For the data Weev encountered was there any possible way to get a 401 response for the URLs?

Sure there was. If AT&T had configured their server properly then that's exactly what it would have given him. If I wanted to introduce some irony then I would have to ask you whether you were "blaming the victim" here.

Re:Orin Kerr: Why I Am Representing Auernheimer Pro Bono on ... by flyinRyan on Friday March 22, 2013 @10:56AM (#5423270)
I concede that it's a bit fuzzy at the moment, but my criteria would be that if a spider could have accidentally crawled this info then it can't be a crime.

>Sure there was. If AT&T had configured their server properly then that's exactly what it would have given him.

That's not a valid test. If a company decides they didn't want you to see something after the fact (as in this case) they can always just claim they didn't configured their servers how they meant to.

>If I wanted to introduce some irony then I would have to ask you whether you were "blaming the victim" here.

What blaming the victim? The victims were the people who's data got released. Since they trusted AT&T with it that would make AT&T responsible. Everyone is talking about Weev but chances are he wasn't the only person on the planet to know about this.

Re:Orin Kerr: Why I Am Representing Auernheimer Pro Bono on ... by AnthonyMouse on Friday March 22, 2013 @11:21AM (#5423474)
>I concede that it's a bit fuzzy at the moment, but my criteria would be that if a spider could have accidentally crawled this info then it can't be a crime.

I don't think that works as a test either. Spiders index whatever other websites link to. The URLs may have been trivial but if there were no public links to them then a spider wouldn't have followed them. And then on the other hand they would be on another website if anyone (like weev) had linked to them, which you can do just as easily with a link that will cause a buffer overrun, and the spider will then follow it and overrun the buffer. It's completely plausible for a search engine to provide you with a search hit which if you click on it will cause a buffer overflow on the destination server and give you root access to the machine, because some "hacker" posted such a link on their site and the search engine indexed it and put it in the database.

>That's not a valid test. If a company decides they didn't want you to see something after the fact (as in this case) they can always just claim they didn't configured their servers how they meant to.

That's what I'm saying. All prosecutions for "unauthorized access" are like that, because if the server had been configured properly then unauthorized access would be impossible, so when it's discovered that it was misconfigured after the fact, the server operator wants to go back and retroactively label the conduct as unauthorized even though their computer allowed it.

There is certainly a matter of degree as to how far you had to go out of your way to get the server to do something you want it to do, but that is such a hopelessly vague and meaningless line between legal and illegal actions that (as Prof. Kerr has argued) it's potentially unconstitutional, to say nothing of whether it makes for good policy.

>What blaming the victim? The victims were the people who's data got released.

Again, that's the point. The law is stupid. The culpable party here is AT&T for putting its customers' info at risk. The party being imprisoned is the one who publicized the vulnerability rather than the ones responsible for putting it into production. I don't know if I support actual criminal penalties just for operating a vulnerable server, but I certainly take issue with the idea that if you do that and then someone publicizes your incompetence, you should have the right to put them in prison for it based on some vague notion of having gone too far in proving the point.

Re:Orin Kerr: Why I Am Representing Auernheimer Pro Bono on ... by flyinRyan on Friday March 22, 2013 @12:56PM (#5424104)
Very well, I guess we're probably in violent agreement.
Re:Orin Kerr: Why I Am Representing Auernheimer Pro Bono on ... by ScottBurson on Friday March 22, 2013 @03:50PM (#5425215)
See EF Cultural Travel BV v. Zefer, 318 F.3d 58 (1st Cir. 2003) (the fact that a website owner “would dislike” the use of an automated script “to construct a database” of information available from visiting the website does not render the use of the automated script an unauthorized access under the CFAA).

Interesting that a court has already opined this. I don't see any other basis on which one can define what constitutes unauthorized access, than whether the website owner would dislike it. I think, as you have argued, that a court looking for some more technical definition will find there isn't any that holds up under scrutiny.