Gmail Messages Are Vulnerable To Interception

Michael Wally writes "GMail messages are vulnerable to interception. An attacker has only to transmit malformed test messages to himself, and information left over in memory, from previous messages destined for other people, will appear with the test messages, in the attacker's inbox. Sometimes, this information may include usernames and passwords... Do you use GMail? Are your communications private? Should they be? Well, here's what we figured out about the issue, that may or may not help you - or perhaps GMail, if anyone can get ahold of their developers, to tell them about it." Update: 01/12 22:21 GMT by T : Good news for Gmail users; those malformed messages are no longer being accepted; read below for a message from Chris DiBona.

chrisd writes "Just so you know, at 10:15am PST mails with the problematic formatting as described in your previous story stopped being accepted into Gmail. Previous emails that had this problem will also no longer will be accessible. If you don't mind, I'd like to take the time to remind Slashdot readers that they can send bugs that may have a security aspect into [e-mail]. If they like, they should feel free to cc me at [e-mail]. We appreciate your patience and we're sorry about the bug."

Wow

[bperkins (12056)]

Did any of this "left over" information happen to be spurious commas?

Re:Wow

[TedCheshireAcad (311748)]

,,,, no, ret,u,rn to yo,ur work. ,,
,do,not,,worry abou,t t,he com,mas.

Broken XML

[Glonoinha (587375)]

Jesus - am I the only one to recognize this bug?
This is just the most publicly seen instance but broken XML does this every single day.

Use the greater than and less than signs as data delimiters in the 'next generation' of data encoding (XML)? WTF were they thinking?

I'm not 100% they are using true XML but from the looks of it if they aren't they are using a home-built XML wanna-be and - well it looks like I was right a few years ago when I (unsuccessfully) campaigned against doing it that way. Not that I campaigned very loud, as I am basically a nobody.

Not broken XML at all

[JimDabell (42870)]

This is just the most publicly seen instance but broken XML does this every single day.

XML never does this. XML parsers, upon finding a problem must stop parsing and throw a fatal error. It's in the specification.

Instead of mindlessly knee-jerking because you don't like XML, try reading the article. The greater-than symbol that causes problems is the delimiter for the email address - syntax that goes back to 1982's RFC 822 - long before XML's time.

Security Category in Gmail Bugs List?

[dolo666 (195584)]

Is it just me or do you find it strange that in the list of known Gmail bugs [google.com], there is no catagory for Security? I'm trying to find out if this bug is one of the known bugs, but I'm guessing it's not? And I'm also guessing that Security is not a concern for Google at this point, which is a very bad thing, IMHO. People are relying on Gmail because of its awesome features, but if someone can read insecured data directly from memory, it's a really big problem -- perhaps even a global design flaw of the system. No wonder Google plays their cards so close to their chest... I just hope they take some amazing measures to prevent these types of bugs in the future... like when somone does >>> or >>>> etc...

I use Gmail and this bug sort of disturbs me. Aren't they using a proper preg check to see if the fields are enclosed with < > ? I'm not even sure how this bug could exist in any normal computing system. I guess the gmail system is a hybrid of some kind? This is indeed very telling...

But it doesn't make me want to stop using Gmail. It's a random security breech that looks like they could fix it in an hour if they wanted to. Time to stop checking my email for a while until this is fixed...

Re:Security Category in Gmail Bugs List?

[TrippTDF (513419)]

This is just a shot in the dark, but I'm willing to bet Google left Security off the list on purpose. a security flaw becomes a lot harder to exploit if the general public does not know it is there.

I don't hold this against Google at all. I'm glad they are not telling the world how to break into my account...

or rather

[apparently (756613)]

#include <std/security_through_obscurity_rant.h hey moms, it's big poppa here! be looking to fly with you 2nite an' get a little stank on mah hanglow, dig-it?! It's gonna be a <B

Re:Security Category in Gmail Bugs List?

[isomeme (177414)]

People are always saying that, but it just isn't true. Relying only on obscurity for security is probably a bad idea, but as part of a complete security solution, it can be very helpful.

People will not successfully exploit a vulnerability they do not know about, or attack a system they do not know is there. Even if some fraction of people are in the know, you've reduced your potential attacker count by the fraction of them who are not in the know.

Re:Security Category in Gmail Bugs List?

[Q2Serpent (216415)]

People will not successfully exploit a vulnerability they do not know about

You did read the article, yes? This is exactly what happened.

Re:Security Category in Gmail Bugs List?

[isomeme (177414)]

The discoverers accidentally found a vulnerability. That sort of thing happens routinely. My point regards the wisdom of not widely advertising that vulnerability, on the theory that others are unlikely to encounter it on their own.

If you discover that I've left my car unlocked, I would much prefer that you not festoon it with a large orange banner saying "THIS CAR IS UNLOCKED".

You Win An Award

[rho (6063)]

Most Humorously Appropriate Usage of the Word "Festoon" In A Slashdot Post.

Re:Security Category in Gmail Bugs List?

[innocent_white_lamb (151825)]

That depends entirely on the context.

The lock on a vault generally relies entirely on obscurity to obtain its security. You can't see how the cams are turning inside of the lock so you can't open it unless you know the combination. If you do know the combination, you can open the lock within a minute or so. If someone invents magic X-ray eye glasses that could see through the steel, then the standard mechanical combination lock would be useless.

The question at that point becomes how likely is it that

Re:Security Category in Gmail Bugs List?

[mattgreen (701203)]

There's nothing wrong with deciding to not publicize known security holes while you try to fix them.

Unless you're Microsoft, of course.

The sense of security coming from using a beta?

[Behrooz (302401)]

The sense of security coming from using a non-publicly-available product that is still in beta? Where the banner "Gmail by Google - Beta" is displayed at the top left of every page loaded? Where the 'Security' section of the user agreement is:

Security

You must promptly notify Google of any breach of security related to the Services, including but not limited to unauthorized use of your password or account. To help ensure the security of your password or account, please sign out from your account at the end of each session.

Oh yes, Google is certainly lulling us into a false sense of security.

Re:Security Category in Gmail Bugs List?

[Basje (26968)]

I'm all for full disclosure in public software. But gmail and other web services aren't public software.

Full disclosure has a purpose: to educate users/admins in order to prevent damage to them. It should not be goal in itself.

In case of proprietary software running on a machine nobody but the developer has access to, why bother. It's not as if the users run more risk if FD isn't practiced. Au contraire.

The only reason I can think of that would warrant FD, is when you want to keep tabs on the developer,

Re:Security Category in Gmail Bugs List?

[Eric Giguere (42863)]

I'm not even sure how this bug could exist in any normal computing system.

It happens the same way that many (most?) bugs happen -- the human programmer forgot to check for boundary conditions in the data interpretation. As the old saying goes, "garbage in, garbage out" -- if you don't validate your data, you may be surprised at the results you'll get. Here the result is that it's exposing someone else's message to you. But it's not that surprising.

These things usually boil down to human error and inco

Re:Security Category in Gmail Bugs List?

[PhilHibbs (4537)]

What do you mean, "I'm not even sure how this bug could exist in any normal computing system"? Buffer overruns are everywhere. Although the classic buffer overrun involves getting the app to write beyond the buffer's bounds and into the stack, this one is getting it to read beyond the point that it should. Unless the system has memory protection built in (and that is only possible on very recent processors) then this is entirely unsurprising. "Some kind of hybrid"? You're not making sense.

One Key Word

[wcitechnologies (836709)]

BETA. It is unlike google to release half-assed web services. Keep in mind GMail is still being offered as a preview, you can't even sign up unless you know somebody else who has it.

Google will work out the kinks, they always do.

Re:One Key Word

[by Anonymous Coward]

Next up on Slashdot: the Google apologists vs. the Apple apologists in a brown nose-off...

Re:One Key Word

[Richie1984 (841487)]

Even though GMail is still being offered as a preview, there seem to be more invites then there are people willing to sign up, in my experience. Basically, in my opinion, if you want an account it isn't too hard to get one nowadays

And while GMail is still in Beta, it is still a widespread and widely used email service. So, while I can understand that there are still bugs in the service which Gmail could iron out without too much trouble, I would disagree with people who underestimate the severity of tho

Re:One Key Word

[MightyMartian (840721)]

> Google will work out the kinks, they always do.

Let me know when they fix the disaster known as Google Groups 2. They've buggered up a ton of archive references, and don't exactly seem to responding in a stellar fashion to the problems.

Re:One Key Word

[the_mad_poster (640772)]

Agreed.

Not only that, as always, e-mail from one network to another across unknown intermediaries is not private. It travels on public wires across public networks. If there's a value in someone targetting you and you're not technically competent enough to know you shouldn't use gmail for important discussions, they can just snap a packet sniffer onto your gateway and watch everything you send and receive right at the source with little fuss and no muss.

First thing's first: you ought not be relying on gen

Re:One Key Word

[phats garage (760661)]

I disagree.

I know that its everyones darling, Google, but its not any less of a privacy spilling bug. Look at everyone who jumped on gmail already. Look at the bug itself, their servers trust the email client to terminate a string.

Never trust an internet client to provide properly formatted strings. Google blew it. (Besides, they're on my bad list for screwing up the usenet archives anyways, they're turning evil.)

Beta..

[ackthpt (218170)]

Beta...beta... Golly, I wonder what that means.

Oh, sure, it means ready to be shipped/used in production by some companies, but has that line gotten to fuzzy for some people?

"that's not a feature, that's a bug"

Re:Beta..

[ad0gg (594412)]

news.google.com has been in beta for 3 years now. Same with google groups, same with froogle. Pretty much the only thing that google hasn't labeled beta is their main search engine.

Email isn't secure

[krog (25663)]

and should never be treated as such. If you want security, use strong encryption.

This is as it was 10 years ago, 5 years ago, now, and in the future. Plaintext should be treated as though you were sending a postcard in the mail.

A Darker Shade of Grey Hat

[American AC in Paris (230456)]

When you find a bug like this, you should first and foremost submit it to the party responsible for the maintenance of the code. You should at least give the responsible party the opportunity to review/respond/repair before making vulnerabilities public knowledge.

Security exploits are a serious matter, and they need to be handled properly. Throwing this kind of thing out in the open willy-nilly is, at best, irresponsible. For one, it means that Google must now rush a fix for something which may have already been in the bugfix queue; rush jobs can disrupt the entire project and increases the odds of human error--which can lead to unnecessary security vulnerabilities.

As for these guys getting hired by Google--being smarmy twits about Google's code review practices probably isn't gonna help their case any. Shame, because a little tact and professional courtesy would have given them a damn good running start at it...

Re:A Darker Shade of Grey Hat

[argel (83930)]

What's this - the 10 commandments? You seem rather opinionated. Clearly the guys concerned don't agree with you. [...] They've spotted a bug in beta code and decided it was easier to tell the public rather than Google. Good luck to them.

Because it has become standard practice in the industry to inform the vendor and give them a reasonable amount of time to come out with a patch before publically annoucing the exploit. It's called professionalism a.k.a. an endangered species here at slashdot.

Re:A Darker Shade of Grey Hat

[pthisis (27352)]

Because it has become standard practice in the industry to inform the vendor and give them a reasonable amount of time to come out with a patch before publically annoucing the exploit.

Key here is "reasonable amount of time", which should be no more than a couple of weeks. Even that may be too long and many vendors will threaten you with lawsuits for going public once you've privately informed them of security holes.

As Bruce Schneier (author of Applied Cryptography, creator of Blowfish/Twofish, etc) writes:

What we've learned during the past eight or so years is that full disclosure helps much more than it hurts. Since full disclosure has become the norm, the computer industry has transformed itself from a group of companies that ignores security and belittles vulnerabilities into one that fixes vulnerabilities as quickly as possible.

Note that Schneier does say:

I believe in giving the vendor advance notice. CERT took this to an extreme, sometimes giving the vendor years to fix the problem. I'd like to see the researcher tell the vendor that he will publish the vulnerability in a few weeks, and then stick to that promise.

Also from the same article:
http://www.schneier.com/crypto-gram-0111.html [schneier.com]

During the early years of computers and networks, bug secrecy was the norm. When users and researchers found vulnerabilities in a software product, they would quietly alert the vendor. In theory, the vendor would then fix the vulnerability...There were incidents of vendors threatening researchers if they made their findings public, and smear campaigns against researchers who announced the existence of vulnerabilities (even if they omitted details). And so many vulnerabilities remained unfixed for years.

The full disclosure movement was born out of frustration with this process. Once a vulnerability is published, public pressures give vendors a strong incentive to fix the problem quickly. For the most part, this has worked. Today, many researchers publish vulnerabilities they discover on mailing lists such as Bugtraq. The press writes about the vulnerabilities in the computer magazines. The vendors scramble to patch these vulnerabilities as soon as they are publicized, so they can write their own press releases about how quickly and thoroughly they fixed things. The full disclosure movement is improving Internet security.

Well...

[grub (11606)]

Yeah, it's a potential privacy breach. That said, using a web-based email system for top secret or potentially embarassing mail is pretty dumb. You get what you pay for, gmail is no different. (nb: I'm a happy gmail user)

Are you communications private?

[skyshock21 (764958)]

Are you communications private?

I don't even know where to start with this one!!!! Editors? You out there???

Re:Are you communications private?

[American AC in Paris (230456)]

I don't even know where to start with this one!!!!>

Simple.

All you communications are belong to them.

Newsflash

[hackstraw (262471)]

Speaking loudly in a public place can be intercepted!

Although this appears to be a valid bug in GMail (that is still beta mind you, and will probably be fixed very quickly), who in the world considers plain text communication secure?

I have no idea who at my ISP has root access (or others that can gain root access) to read my plaintext mailbox.

Nothing to see here... please move along.

Re:Newsflash

[Country_hacker (639557)]

Looks to me like they already fixed it, I tried sending an email without putting the end bracket on the address (Just like the guys in TFA) and it popped an error message. Those guys at Google are on the ball today. :-)

Well hey..

[sinner0423 (687266)]

Google = best & brightest, right?

I mean, their aptitude tests & hiring policies makes me believe they've got a few nobel prize winners working there..

Shouldn't they be able to fix this during lunch break?

You mean there is a server-side bug in GMail

[Idaho (12907)]

From the description, the way you can read messages of other people has nothing to do with 'intercepting' messages. Man in the middle attacks are always possible, but this looks like a simple serverside bug (buffer overflow or string formatting problem, most likely) which will probably be fixed on short notice.

I don't think you can do directed attacks either (e.g. 'intercept' only the mail of a specific target). So I think it's not a real showstopper.

Still, it shows that even Google can make mistakes in their code...who would have thought! ;)

end of the world is coming!!

[jxyama (821091)]

headless $500 Mac and $99 iPod...

now Google messes up...

with all the natural disasters happening, i cannot think of a good reason why the world wouldn't end the day after tomorrow.

Well...

[slavemowgli (585321)]

Serious as it may be, this does not allow you to selectively attack a specific person or account - you just have to "hope for the best", so to speak. While I wouldn't underrate it (is that a word?), I wouldn't overrate it, either, and I'm pretty sure that the Google people will plug this in no time. It's been my experience that they do look at reports that are coming in (just like they claim), and that they are generally quite quick to fix even minor issues, so something that is security-related *and* (by the sounds of it) easily fixable shouldn't last long.

That being said, did the authors actually contact Google about this prior to making the whole thing public? Full disclosure is good, of course, but it's also nice to give the vendor a chance to fix things before you inform every script kiddie in the world about what you found. :)

All email is vulnerable.

[pavon (30274)]

To everyone expressing concern about using gmail in light of this exploit - I hope you know that all email is vulnerable to interception. It is sent as plaintext across the internet, and hops though a dozen servers before ending up at it's final destination. This exploit is just another way to do something that has been possible by design ever since email was created.

If you want your email to be secure you have to encrypt it. Otherwise don't have any expectation for privacy.

Re:All email is vulnerable.

[Carthag (643047)]

This exploit uses a flaw in Google's code that allows viewing of memory on Google's servers. Hardly an inherent flaw in email as such.

GMail vs Hotmail

[kevin_conaway (585204)]

Why is everyone brushing this off by saying "well you should have known that email isnt secure, tough luck!"

If Hotmail had this bug, everyone here would be up in arms.

Just because email isnt secure doesnt mean this isn't serious. I would hate to think of all the people reading my responses to craigslist postings :)

Re:GMail vs Hotmail

[valkraider (611225)]

If you know that your product may well have serious bugs like this, you shouldn't have sent out a press release promoting its launch, you shouldn't have given away free accounts to thousands of Blogger users, you shouldn't allow people to fire off a bunch of invitations to anyone they choose, and you should make some indication on the website (beyond "BETA," not everyone who uses Google reads Slashdot) that there can be risks associated with using it.

Hmm. I wouldn't try Windows if I were you...

Way to go, jerks.

[Canthros (5769)]

You did notify Google and give them a reasonable period to time in which to respond, right? Because you've just shouted, in the loudest possible way, how to access all that data you're so worried about protecting.

SPAM!

[knitterb (103829)]

Chances are, since most email these days are spam, an attacker is going to have to go through a lot of spam before finding something interesting.

Yawn...

[revery (456516)]

I already read about this in a newsletter that I received in the "Reply To" field of an email.

--
Was it the sheep climbing onto the altar, or the cattle lowing to be slain,
or the Son of God hanging dead and bloodied on a cross that told me this was a world condemned, but loved and bought with blood.

This was more about their 15 minutes than Google.

[EvilFrog (559066)]

Many other people have pointed out that GMail is still in beta, and that if they would have told Google first it probably would have gotten quietly fixed without any damage being done.

Of course, they acknowledge that, but they're arguing that they're helping protect people by making them aware of the problem.

I call bullshit. This is about them wanting recognition for finding the bug. If they would have sent it to Google, it would have been fixed and no one would care who discovered it. Because they went public with it they can boast that they were the ones who found the bug.

Of course, it swings both ways. Now if someone uses this exploit and steals your password (which is honestly rather unlikely), you know who to blame for making it public knowledge before Google had the chance to fix it.

Hacker Hubris

[Jtheletter (686279)]

Wow, are these guys full of themselves. I write complex automation code for a living, in an environment that demands rigorous QA practices and documentation, but guess what? We still create bugs, find latent bugs that have gone undiscovered for many builds, and even get some real DUH! headslappers from time to time. Fact of the matter is, when you've got a couple hundred thousand lines of code there are going to be errors and unintended consequences, mostly arising out of missed checks, such as this gmail problem (assuming they're right about the end tag check being the cause).

For these people to find a single issue in such a system, then say it's a shortcoming of gmail's QA process, and in the same breath ask for work - implying they've got the skills to even handle such a job - is insulting. Please, just because you're smart enough to expose a flaw once you stumbled onto it in no way means you are qualified to correct that or any other issue. Sometimes our QA team finds a flaw and even digs in the logs enough to pinpoint the problem but it can still take the developer who designed the code days to correct.

In other words, noticing that you're bleeding does not qualify you as a surgeon. Instead of publishing their finidings in a detailed how-to, these asshats should have forwarded the info to gmail and let them deal with it, and that's assuming that the gmail team didn't already have it in their list of bugs. I just don't understand why people feel the need to not only describe a security problem, but give every hacker on the net a roadmap as to just exactly how to use it and what illicit activity it might be good for.

Did anyone else see this?

[Lank (19922)]

At the bottom of TFA:

Screen Capture #5 [milatic.net]
Jack Rabbit Vibrator Features

This message describes the features of one "Jack Rabbit Vibrator," a 7.5" Multi-Speed toy of sorts.

What are the odds of finding that?

A Job?

[jayloden (806185)]

lots of comments here are noting the hubris of these guys in asking for jobs.

I'd just like to add that not only are they criticizing the company's QA process and releasing the bug without having notified google first, as others pointed out...

They found the exploit by MISTAKE! It was a bug in their own code that caused the problem, something as stupid as a missing caret at the end of a line. So, in other words, they are looking for work looking for bugs in Google's software that they found solely because of a bug in the software they wrote.

On another note, bugs in software happen, no matter WHO you are, the trick is just to be able to fix them in a timely fashion and deal with the situation effectively. I believe that Google will do this, especially if the previous comment stating that it has been patched is true. Everyone is making too big a deal out of something that has happened to every developer on every software ever. The reason MS gets crap for it is simply because they continuously produce buggy code ridden with security issues, but deny this is the case, and often ignore security problems until they are found out by the general public.

-Jay

Looks like GMail is not accepting Mail

[ahsile (187881)]

I'm assuming this is until the problem is fixed:

"APPLICATION" 516 "2005-01-12 20:01:48" "SMTPDeliverer - Message 15213: Delivering message from [e-mail] to [e-mail]."
"TCPIP" 516 "2005-01-12 20:01:48" "DNSResolver - MX Lookup: gmail.com"
"TCPIP" 516 "2005-01-12 20:01:48" "DNSResolver - MX Lookup result for gmail.com: 3 servers"
"APPLICATION" 516 "2005-01-12 20:02:09" "SMTPDeliverer - Message 15213: Failed to connect to gsmtp185.google.com."
"APPLICATION" 516 "2005-01-12 20:02:30" "SMTPDeliverer - Message 15213: Failed to connect to gsmtp171.google.com."
"APPLICATION" 516 "2005-01-12 20:02:51" "SMTPDeliverer - Message 15213: Failed to connect to gsmtp57.google.com."
"APPLICATION" 516 "2005-01-12 20:03:13" "SMTPDeliverer - Message 15213: Failed to connect to gmail.com."
"APPLICATION" 516 "2005-01-12 20:03:13" "SMTPDeliverer - Message 15213: Failed to connect to all [e-mail]'s mail servers."