Slashdot Mirror
GoogHOle Exploits GMail, Picasa and 200K Other Sites
Giorgio Maone writes "Multiple Google-targeted exploits disclosed in the past 3 days could compromise your GMail account, steal your pictures from Picasa or impersonate you on almost 200,000 big sites which outsourced their search engines (vulnerabilities included in the price). If even Google, a very reactive company when web security matters, does face this kind of problems, how serious is the threat and what can you do, as a "normal" web user, to protect yourself?"
How do we blame this on Microsoft?
at the end of the day, when you rely on third party apps run by a completely different company, you can't do ANYTHING to protect yourself.
The article is very low on details. I read it and I'm still not sure how it works, whom it affects and what I can do to protect myself (obviously, since I don't know how it works).
It would have been nice if they went into some more detail for technical users.
Send email from the afterlife! Write your e-will at Dead Man's Switch.
According to the article, exploint uses Cross-site scripting, also known as XSS. There is a firefox plugin called NoScript that limits cross site scripts. The article points you to http://noscript.net/features#xss which describes the anti-XSS protection of noscript. The noscript pages suggests that you only load firefox plugins from addons.mozilla.org and sends you to https://addons.mozilla.org/en-US/firefox/addon/722 where you can download noscript.
--- Often in error; never in doubt!
Don't trust your data to 'on line' providers.
---- Booth was a patriot ----
You'll never be safe.
Complex software designed for diverse interactions will always be vulnerable to some kind of attack, even if it's as simple as someone walking out of a data center with a thumb drive in their pocket. Almost every vulnerability stems from a "feature" implemented to make software easier/flashier/useful. Flexibility and expansiveness carry with them the price of vulnerability, and pretending otherwise is to wear blinders.
Of course developers should do their best to prevent security problems -- but there is only so much that can be done when you also need to implement Really Cool Stuff. Every door you make is a door than can be kicked in, no matter how good your locks. The real world has never offered perfect security because it can't -- why expect engineered items to be safe from all evil?
Treat software and computers with caution, like walking through a major city's downtown at midnight. Sure, it's dangerous at times -- but it can also be exciting. Just don't pretend that danger doesn't exist...
All about me
FTFA:
... but I already use a separate SeaMonkey browser profile for my GMail account (don't want it being associated with my normal Google searches), and access untrusted URLs using another browser running under a different user. As a matter of habit (I do web-based stuff and I'm used to having several different browsers open). Probably not 100% foolproof, but helps me sleep easier at night.
If even Google, a "very reactive" company faces these issues, what can be done? The answer: Nothing can be done.
There is no way (unless you're writing something with hundreds, rather than thousands of lines of code) that every code path is going to be audited carefully enough to catch every possible bug. Good coding practices aside, programmers are human and make errors. You do your best to catch as many as you can, and that's all you can do. When you're a "consumer" of code, you look for an organization that seems to be doing this and use their stuff. There's no complete, proactive solution to bugs.
The important thing is that you want someone "very reactive." An organization that acknowledges these flaws up-front, publicly announces vulnerabilities with a work-around until they're patched, and then corrects problems in a timely manner. Some companies are more like this than others.
Interested in a Flash-based MAME front end? Visit mame.danzbb.com
At the end of the day you can sight all kinds of flaws in Microsoft and closed source software. However, for as you're running that software LOCALLY on your computer, then you have the ability to take measures to protect yourself.
If you're drinking the google-juice just because it's "cool" or you want to support them because they're "not evil", you're only doing yourself a dis-service.
Keep your email local, dont save your passwords on a public "service", dont keep naked pictures of your girlfriend on your "G-Drive", etc etc etc
Common Sense
Neither can you if you hire people to implement it on your own company.
And if you do it yourself, you can be sure that the security will not be higher than your own skill set.
If you want to trust nobody, you might as well retreat to am isolated island somewhere, as you will be unable to function in a society. The key to functioning in a society isn't distrust, but to to be able to judge who to trust and who not to. Which is quite annoyingly mostly a social rather than a technical skill.
----
I personally trust the people at Google more than I trust the people and products responsible for our internal mail solution (which is also available as web mail). Especially with regards to competence (as opposed to integrity). So I would love for us to switch.
perhaps one of the simplest examples of a program involving transactions and user interaction
now consider the number of hacks you can use to exploit a vending machine (granted many are physical hacks, but you could call that analogous to social engineering hacks involving "real" software)
now, if something as simple and as straightforward as a vending machine can be exploited, then the obvious conclusion is that:
we should not express shock that google can be hacked, but we should express shock that any of us expected it couldn't be hacked
any computer program of sufficient complexity will be hacked. not could be. will be
and the internet is well into the zone of "sufficient complexity"
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
Turn off client side scripting.
OR
echo "127.0.0.1 google.com" >> etc/hosts
When I first started in web development it was hammered into us that client side scripting MUST degrade gracefully. What ever happened to that rule?
I hate sites locked to "Web2.0" only! For the most part I will not use them. There are only a handful of URL's in my scripting white list, most of them my own sites.
Yes, I use some client scripting, but it degrades properly.
I don't let websites keep my credit card info, or any password other than the one needed to unlock their own site, or any other personal info that is valid outside their own realm, unless their service won't work otherwise.
The Web would be a lot more secure if my browser had a keyring integrated with my own computer, and I kept my secrets on my own computer under my own control. When challenged by any server for a secret, my browser or other client SW I'm using should pull the secret from the keyring and supply it to the server. That service should let me use a master key from any remote terminal to query my own computer, over my home broadband or wherever I keep the secrets. All by a standard protocol that lets me just fill web forms (and other challenges) as I do now, possibly entering the master key and maybe an additional confirmation challenge to let the 3rd parties communicate, but otherwise just as transparent as just filling in the forms.
If a 3rd party server is going to store my secrets, I want it to be my bank. I don't know why banks haven't gotten into this business already, after well over a decade watching their profits multiply from the Web, along with many risks. Maybe Google will push a key distribution protocol like this in partnership with some banks. That would also finally get Google into the payment business to challenge eBay's PayPal, which I hate precisely because its (mostly unregulated) global Internet bank is a monopoly, and I don't trust PayPal with my secrets. If Google does recover from this crack, they might be solid enough to trust.
--
make install -not war
Of course, exploitable programs are all Microsoft's fault - which must be why the remote root exploits for Quake 1 and 2 for Linux must be all Linus' fault!
Let's be honest, exploitable applications are OS independent. Though I guess honesty never really comes into it with you, hmm?
"It does not do to leave a live dragon out of your calculations, if you live near him." - Tolkien
I see many here making excuses for Google ("You'll never be safe with online service providers", "There's nothing Google can do", etc) and offering solutions ("Use Firefox with Noscript", etc). But I can't help but laugh because I know that if this were about Microsoft web services being exploited, the comments would be completely different. The number of comments would be at least five times greater than they are here and would be filled with gloating and screaming over Microsoft's "incompetence" and whatnot.
You know that there is some truth in what I say.
It looks to me that there are major holes in Google's services, and they need to be called out on it, not given excuses.
-- "I never gave these stories much credence." - HAL 9000
How about people who were looking to move their internal office applications to google (there were hundreds of people here on Slashdot saying they were planning on doing just that), are their critical private documents at risk or not? I've never been fond of software as a service for internal business functions, and this seems like another concern point against it.