The Current State of the Malware/AntiVirus Arms Race

An anonymous reader writes "An article at Net Security explores how malware has developed self-defense techniques. This evolution is the result of the double-edged sword of the malware arms race. Anti-virus technology is ever more advanced, but as a result surviving viruses are increasingly sophisticated. What Net Security offers is a lengthy look at the current state of that arms race. 'There are many different kinds of malware self-defense techniques and these can be classified in a variety of ways. Some of these technologies are meant to bypass antivirus signature databases, while others are meant to hinder analysis of the malicious code. One malicious program may attempt to conceal itself in the system, while another will not waste valuable processor resources on this, choosing instead to search for and counter specific types of antivirus protection. These different tactics can be classified in different ways and put into various categories.'"

Comments

When programers fight...

[Coraon]

it's the computers that suffer. Wont someone please think of the computers?!

Re:When programers fight...

[Dimentox]

WARNING WORM DETECTED: By reading the parent you have been infected with the new slashdot worm. To remove it Please click here

Evolution?

[truthsearch]

Malware evolution? That's just theory and conjecture. If god had wanted our computers to be free of viruses he wouldn't have invented Microsoft.

(There goes some karma.)

Re:Evolution?

[geoffrobinson]

The viruses are intelligently designed. I'm not vouching for Microsoft Windows.

From TFA

[Chris Tucker]

"This article will only examine malicious programs written for the Windows operating system (and its predecessor, DOS) due to the rarity and relatively small number of malicious programs for other platforms."

OK, you had to go to the second page of TFA to see this, but at least they came right out and said that Windows is the primary and almost exclusive target of malware.

Unlike almost every other article about viruses and malware in recent years.

Mac OS X: Because it was easier to make UNIX user friendly than it was to fix Windows!

Re:From TFA

[kebes]

Market share is certainly a factor, but I think it's a stretch to say that it's the only factor.

Let's say some nefarious guys are trying to get their malware installed on everyone's computers. So they buy some exploit code that targets IE. They say "Great, this will infect 3/4 of the computers out there!"

Now if these malware distributors are approached by some other guy who says "I can sell you exploit code that targets Firefox"... do you think the malware distributors will say "no thanks" or will they say "Great, that covers the other 1/4 of computers out there!" (Maybe they will pay less for that exploit, but they will surely use it if it's available.)

Since Firefox's market share is not insignificant (10% to 25%?), there should be a market for such exploits. Similarly, there should be a market (perhaps smaller, but still a market) for the 4% Mac users. It appears that despite this, the targeting of Mac and Firefox is very much less than Windows/IE (more than can be accounted for by market share alone).

I'm sure that part of it has to do with market share. However inherent security is also part of the equation. (And frankly I don't know why such a statement is so controversial on Slashdot... why should security be based on only one factor in the first place?)

Re:From TFA

[Opportunist]

Security is by definition the minimum of the system's capabilities and the user's. When the system can't hold its water (or data), the user can be the best security guru in the world and it is insecure. Likewise, the system can be as tight as possible, with a clickmonkey at the helm it is hopeless (provided it's an all purpose machine that doesn't restrict the user's ability to cause havoc).

Still, market share is a key factor when it comes to malware. Malware "kits" cost a wee bit of money, ranging from a few hundred to a few thousand USD, depending on sophistication and "additional services" (let's not get into too much detail, you get the idea). Basically, everyone develops for IE on a WinNT core machine. Why? Market.

Yes, there would be a market for FF exploits. But it's smaller. Development costs are pretty much equal for FF and IE exploits, and you can not really develop a "generic" exploit that targets both, unless you target the OS underneath and not the browser itself (that happens too, but generally requires a lot more knowledge about the OS itself, and it is by far less flexible). Since the cost of spreading malware is roughly equal for whatever you want to land, and doing so is not really cheap, attackers usually try to maximize their efficiency by limiting themselves to the most popular OS/browser combination (provided they want to do ID theft attacks). At the very least, they will limit themselves to the most popular OS.

The limiting factor here isn't that the "kit" itself would be costy. Yes, you might have a FF exploit kit available and you'd sell it for a fraction of the IE kit (but why should you, you could more easily develop an exploit kit for IE (there are effing templates for it in VC!) and cash in). But the spreading cost for either malware stays the same.

Thus the usual exploit targets IE/WinXP. Should the market share of FF rise, I'd wager to about 35-40%, we'll probably see mass spam of FF targeted malware, due to people using FF feeling secure and are thus maybe less wary. It might happen. But generally, you'll never see masses of malware for non-mainstream targets (OS, browser, webserver...). The cost of spreading is the same, no matter what your target is. So why shoot at something but the biggest target?

Lack of Specifity on Infection Vectors

[Cr0w T. Trollbot]

My brief overview of the article leads me to believe that it's long on general malware theory, and short on the specifics of current malwear infection vectors as opposed to techniques. I believe that most of the readers of Slashdot are familiar with how a rootkit works. Far more valuable would be a breakdown of the most common infection vectors for rootkits right now. Is it TCP/IP stack overflows, Active-X controls, e-mail trojans, or old-fashioned human error?

Fruthermore, "trends" in malware construction obscure the reality that certain software packages (Windows, IIS) are otrders of magnitude more vulnerable than others (OS X, Linux, Apache). The unstated elephant in the room is that 95-99% of malware attacks are due to Microsoft vulnerabilities.

Crow T. Trollbot

Re:Viruses will never go away

[doti]

And how will they compete with Free software anti-virus?

Oh please...

[Opportunist]

This conspiracy is about as old as the AV industry. At least you spared us this time the drivel about AV vendors first of all creating malware so they can sell their stuff.

Basically it's impossible to write the perfect AV software. It simply does not work. The perfect AV software could, of course, exist: Simply disallowing ANY kind of user interaction and installation of additional products. Perfect computer. Useless, but perfectly safe.

The problem is that malware does not use anything "special" that makes it easy to say "something that uses function X or accesses Y is malware". Doesn't work that way. What malware does it usually not much different from normal program activity. They access the windows registry, create keys there, they create and alter files (not necessarily system files, which would be "suspicious" behaviour to say the least), they plug into Internet Explorer, they open ports for incoming connections, they transfer data to and from the computer.

It's not anything that is by defintion "bad". How'd you want to create the "perfect" AV product?

Re:Viruses will never go away

[42Penguins]

and I will yawn and enjoy a relatively un-penetrated life.
That's pretty much a given here on /.

Re:Viruses will never go away

[gardyloo]

This XP install has been going for over a year

    Geez, and I thought Gentoo was supposed to take a while.

And the collateral damage of this' war'...

[rickb928]

...extends beyond poor performance, spam, cost of software, etc.

We got hit here with a collateral listing of one of our tools as 'spyware'.. It shut down our software across the U.S.

We used a toolkit from a vendor to encrypt and compress files for transmission and for patch distribution. It was slick, lightweight, and sufficiently secure. it was also a commercial product, and was sold to another publisher who used it in their software.

One of their packages is an IM logging and monitoring tool. Good for AOL IM, and others. You have to either download it as shareware, or buy it outright, and then you have to install it, with the usual requirement that you actually have access to the PC. It's not and has never been distributed as 'spyware' in the sense of an unexpected or unsolicited install, nor was it ever distributed from a website or as part of another package - unless you repackaged it yourself. The biggest users were corporate IT departments monitoring IMs for compliance, and parents/spouses/etc snooping on others.

Not what I think of as 'spyware'. But someone else thought differently.

The IM logger got reported to either Trend Micro or McAfee as 'spyware' more than a year ago. Sporadic reports continued, until the latest (?) release came out and got popular. Then the flood of reports ensued. And when I say 'flood', I mean 'dozens'. I suspect some HijackThis logs started showing it, and after a few more reports, it was assumed by someone that this application was part of other kits. Listing the application by one anti- company leads to everyone else listing it. No one wants to be left behind, and none of the 'security' companies wants to be the one that lets bad stuff in, just because they actually evaluated the listing. No, it got listed by everyone.

And the controls along with it. Including the one we used for everyday, legitimate encryption and compression.

Our customers started reporting failed installs and reinstallations. One reported they got a virus alert. We looked things over. Why now? We hadn't changed anything substantial in years.

Then, on a whim, I Googled for it. BAM! Our control was listed as malware. WHA?

We figured it out an an hour. I asked around some of the contacts I knew at Symantec, etc. Their advice was simple - give up. Go get a new tool, recode, and move on. Surrender. Even though the module we used was by itself harmless, it was guilty by association. So we did. So far as I know, the company that produced these tools & modules is struggling with this. After all, their code signatures are now officially 'malware'. Kinda like banning drills 'cause someone drilled a hole in their finger by accident. Pretty soon, nothing gets drilled. Not a good state of affairs for the drillmakers.

And not a good state of affairs for drill users, either.

That IM logger that started all this? It was commercial software, and other than being highly annoying for kids who value hiding their IMS from snooping parents ("Hey, who's paying the Internet bill around here?"), or spouses caught on dating sites, the businesses forced by law to treat IMs as if they were business correspondence found this to be a good tool. Not so good any more. About the only way to use this is to keep writing exceptions to your anti- software. If you can. And keep re-writing these exceptions every damned update. Maybe more than twice a day.

It looks like this application is dead. Kinda sad.

We survived, though some of our customers did get concerned. In our business, being labelled as 'spyware' could cause massive problems, beyond the usual. It could be front-page of the fishwrap stuff.

In the midst of the virus/spyware/malware/anti- battle, this is one small story of how unintended consequences have real costs. We had to scurry to buy new stuff, re-code, and distribute. Our original tool vendor has had to give up on a good product, through no fault of their own. The application vendor that 'st

A question about diminishing returns

[Bombula]

At what point is it simply not worth the effort to write a new virus?

I assume it's getting more and more difficult to write viruses as time goes by - is that correct? If this is indeed an arms race, then one side or the other is going to run out of time and energy and money sooner or later, and I'm guessing it won't be the AV companies since there's so much at stake.

Re:A question about diminishing returns

[Opportunist]

Actually, surprisingly it's getting easier. Think game development. With the advent of DirectX, you needn't know too much math anymore to get some cool looking 3D graphics on the screen. The same applies to malware. Back in the good ol' days of DOS, you had to know quite a bit about the inner workings of the system to get your virus in. You had to redirect software resets, trap a few interrupts, essentially you had to write a driver. Today, most of the malware that circulates could be written in VB. Some is.

Obfuscation is also easier than ever, with a lot of runtime packers and scramblers existing. It's easy to repack a file in batch mode that ensures that no two samples an AV company could get are the same. Thus the simple "signature" approach someone suggested earlier won't find a thing anymore.

It's also not getting trickier to hook into the system. Since there are still the majority of crates running with users having admin access, the same ol' tactics that worked 5 years ago still work. It's also not simple to track the use of "suspicious" calls, since Windows itself makes quite liberal use of functions that e.g. hook keyboard input or inject code into other processes.

Writing malware is also no longer the pastime of bored adolescent geeks. It's business. We're talking organized crime cartels here and that a "virtual" bank robbery (by hijacking online banking sessions) is more profitable and less risky than the real counterpart is a given. When I see the figures, I sometimes wonder why I stay on this side of the fence...

It still is an arms race, but with the AV companies in the defense. Constantly. An AV company can only react to a development, anticipation is pretty much impossible. There are far too many roads the next attack can come from that it's not feasible to develop in a certain direction without anything warranting it.

A few years ago, malware authors started to obfuscate their code. AV companies reacted by developing ways to crack that obfuscation. Then malware attacked certain AV software directly, as mentioned in TFA. The software was adapted to thwart such attempts. Malware started to contain rootkit functionality to hide itself. AV tools started to come with their own file system drivers to read the HD directly instead of relying on system calls.

You cannot anticipate that sensibly. What will be next? I don't know. I can only see trends and development in the malware that runs through my fingers. Which is a very tiny amount of the malware that gets written every day. It's a bit like trying to sieve a beach with a toy sieve. The big thing in malware today is (and has been for about a year or two) remote controlling, setting up servers somewhere and making the malware phone home. Yes, it's no longer IRC. It's a server in Belarus, Kazakhstan or Brazil (or some other country where the police has better things to do than being bothered by a server that doesn't really do any damage in their own country). So some malware packages started implementing tools that can monitor traffic and find "suspicious" traffic, just in case they can't find the corresponding malware. Possibly because the malware itself doesn't exist anymore, it was only an installer that manipulated some system file in such a way to send that info... and so on.

The current thing is (aside of what's been here for ages) id theft. Your amazon or your ebay account, your online banking information, your credit card information, and of course your machine, as a place to spew malware from, as a spambot or simply as a relay to route traffic through to obfuscate the real destination. With broadband becoming the norm and computers running 24/7 to download .torrents, they turn into the ideal dead drop.

There's much at stake. For both sides. I don't see a winner on either side too soon. Well, it's good for my job security, that's a given, but I didn't go into this venue just to make money (it's not THAT well paid). If I wanted that, I'd have learned ABAP.

ClamAV

[DrYak]

And how will they compete with Free software anti-virus?

Actually, by cheating ;-)

Funny little anecdote in the world of virus scanning (harmless although dishonnest).

CalmAV is such an open-source virus engine (with ClamWin as a Windows port).

There have been several studies done about it (links on ClamAV's site) which reported that ClamAV, despite not being a non-commercial project, has among the fastest response time when new threats emerges.

The studies also surprisingly uncovered a small cheating : some companies did small update that didn't bump up the signature release number, but that included the new virus detection. Normally such non-upped releases should be reserved for modification of the sig library that don't affect the number of detected viruses (like repacking the data more efficiently or whatever). But the companies nonetheless tried to slip in newer sigs, hoping that users would not notice it. When doing a retrospective study, unsuspecting users will read that virus XYZ is detected since Sig-file release A.B.C and they will see that Sig-file release A.B.C was released on YYYY-MM-DD HH:mm, thus will come to the conclusion that the virus was detected earlier than the concurrene. (Source, paragraph A dirty little secret).

But anecdote aside, ClamAV is a nice anti-virus engine, that has plugins (either bundled in or 3rd party) that enables on-the fly scanning of data at usual entry points (ClamAV is popular for mail filters in Unix. ClamWin has plugins for mail clients and FireFox's downloader, etc.) and is a nice stuff to put in the "post-download script" of your usual peer-2-peer software. Please note that ClamWin still lacks a on-access scanning mode (although some 3rd party application like Winpooch can start scanner before executing or reading files).