Testing Anti Virus Software Against Packed Malware

Very interesting idea as packed malware is something rather common these days, and as we've seen the recent use of commercial packers in the "skype trojan" malware authors are definitely aware of the concept. What the authors did was to pack the following malware using 21 different packers/software protectors - Backdoor.Win32.BO_Installer, Email-Worm.Win32.Bagle, Email-Worm.Win32.Menger, Email-Worm.Win32.Naked, Email-Worm.Win32.Swen, Worm.Win32.AimVen, Trojan-PSW.Win32.Avisa, Trojan-Clicker.Win32.Getfound, and scan them with various anti virus software to measure which ones excel at detecting packed malware. What some vendors are best at detecting others doesn't have a clue about, but the more data to back up your personal experience, the better for your decision-making.

Threats of Using Outsourced Software

Self-efficiency in (quality) software programming for security reasons -- yeah, sure :

"The possibility that programmers might hide Trojan horses, trapdoors and other malware inside the code they write is hardly a new concern. But the DSB will say in its report that three forces — the greater complexity of systems, their increased connectivity and the globalization of the software industry — have combined to make the malware threat increasingly acute for the DOD. "This is a very big deal," said Paul Strassmann, a professor at George Mason University in Fairfax, Va., and a former CIO at the Pentagon. "The fundamental issue is that one day, under conditions where we will badly need communications, we will have a denial of service and have billion-dollar weapons unable to function."

The billion-dollar weapons system will be unable to function in case of an ELINT attack, not a software backdoor taking the statistical approach.

There's an important point to keep in mind, during WWII, the U.S attacted Europe's brightest minds who later on set the foundations for the U.S becoming a super power. Still, you cannot expect to produce everything on your own, and even hope of being more efficient in producing a certain product in the way someone who specialized into doing this, can. Start from the basics, what type of OS does your Intelligence angency use in order not to have to build a new one and train everyone to use it efficiently? Say it with me.. Moreover, the sound module in your OS has as a matter of fact already been outsourced to somewhere else, if you try to control the process with security in mind, vendors will cut profit margin sales, as they will have to pay more for the module, will increase prices slowing down innovation. But of course it will give someone a very false feeling of security.

Fears due to outsourced software? Try budgeting with the secondary audits "back home" if truly paranoid and want to remain cost-effective. While it may be logically more suitable to assume "coded back home means greater security and less risk", you'll be totally wrong. All organizations across the world connect using standart protocols, and similar operating systems, making them all vulnerable to a single threats of what represent today's network specific attacks. And no one is re-inventing the OSI model either.

You can also consider another task force, one that will come up with layered disinformation channel tactics when they find out such a backdoor, as detecting one and simply removing it on such systems would be too impulsive to mention.

Who's Who on Information and Network Security in Europe

A very handy summary of Europe's infosec entities and contact details that come as a roadmap for possible partnerships or analyst's research :

"This Directory serves as the “Yellow pages” of Network and Information Security in Europe. As such, it is a powerful tool in everyday life of all European stakeholders and actors in Network and Information Security (NIS). By having access to all contact data and entry points for all European actors in one booklet, available on your desk, the “arm length’s rule” of access to information is becoming concrete. I am confident that this device of compiled Network and Information Security stakeholders, contacts, websites, areas of responsibility/activity of national and European Authorities, including organisations acting in Network Security and Information, serves our mission to enhance the NIS security levels in Europe well."

Compared to China's information security market on which I've blogged in a previous post, Europe's R&D efforts are still largely de-centralized on a country level, but hopefully, with the ongoing initiatives among member states innovation will prevail over bureaucracy.

The Zero Day Vulnerabilities Cash Bubble

The WMF was reportedly sold for $4000, a Vista zero day was available for sale at $50,000, and now private vulnerability brokers claim that they beat both the underground and the current incentive programs, while selling vulnerabilities in between $75,000 - $120,000.

"The co-founder of security group Secure Network Operations Software (SNOSoft), Desautels has claimed to have brokered a number of deals between researchers and private firms--as well as the odd government agency--for information on critical flaws in software. Last week, he bluntly told members of SecurityFocus's BugTraq mailing list and the Full-Disclosure mailing list that he could sell significant flaw research, in many cases, for more than $75,000. "I've seen these exploits sell for as much as $120,000," Desautels told SecurityFocus in an online interview."

But the cash bubble is rather interesting. Zero day vulnerabilities are an over-hyped commodity and paying to get yourself protected from one, means you'll be still exposed to the next one while you could have been dealing with far more risky aspects of protecting your network, or customers. The (legitimate) business model breaks when every vendor starts offering a "bounty" for vulnerabilities while disintermediating the current infomediaries. It would be definitely more cost-effective for them, than improving someone's profit margins. Or they could really reboot their position in this situation by applying some fuzz logic on their own software at the first place.

Attack of the SEO Bots on the .EDU Domain

A university's Internet presence often results in very high pageranks for their site, therefore, if a malicious spammer would like to harness the possibilities of having the spammed message appear among the top 20 search results, he'd figure out a way to post direct http:// links on various .edu domains, especially on the wikis residing there. That's the case with PuppetID : Matias Colins -- of course collins is spelled with one L only --. Matias Colins is an automated attack script that's already hosting hundreds of spam pages on the .edu domain, mostly adult related, and it's worth mentioning that where access to a directory has been in place, the hosted pages blocked caching from any search engine, or hosted one on its own. Redirection is perhaps what the attacker is very interested in too. See how this berkeley.edu link - dream.sims.berkeley.edu/~tdennis/wp-content/animalsex.php - redirects to a site for whatever the page title says, and this is yet another one - oit.pdx.edu/jethrotest/mysqldb.php.

Here are two more examples of another bot using my blog post titles to generate subdomains or the like, and of bots abusing Ebay's reputation system by self-recommending themselves.

Social Engineering and Malware

With all the buzz over the "Storm Worm" -- here's a frontal PR attack among vendors -- it is almost unbelievable how hungry for a ground breaking event, the mainstream media is. And it's not even a worm. If you are to report each and every outbreak not differentiating itself even with a byte from previous "event-based" malware attacks, what follows is a flood of biased speculations -- too much unnecessary attention to current trends and no attention to emerging ones. With pre-defined subjects, static file names, one level based propagation vector, with the need for the end user to OPEN AN .EXE ATTACHMENT FROM AN UNKNOWN SOURCE, and with "the" Full_Movie.exe in 35kb, worldwide scale attacks such as the ones described here, are more of a PR strategy -- malware with multiple propagation vectors has the longest lifecycle, as by diversifying it's improving its chances of penetration. Don't misunderstand me, protecting the end user from himself is a necessity, but overhyping this simple malware doesn't really impress anyone with a decent honeyfarm out there. It doesn't really matter how aggressively it's getting spamed, what matters the ease to filter and enjoying the effective rules you've applied. No signatures needed. As a matter of fact I haven't seen a corporate email environment that's allowing incoming executable files in years, especially anything in between 0-50kb, have you? My point is that, the end user seems to be the target for this attack, since from an attacker's perspective, you have a higher chance of success if you try to infect someone who doesn't really know whether his AV is running, or cannot recall the last time an update was done to at least mitigate the risk of infection. These are the real Spam Kings.

At the beginning of 2006, I discussed the evolving concept of localizing malware attacks :

"By localization of malware, I mean social engineering attacks, use of spelling and grammar free native language catches, IP Geolocation, in both when it comes to future or current segmented attacks/reports on a national, or city level. We are already seeing localization of phishing and have been seeing it in spam for quite some time as well. The “best” phish attack to be achieved in that case would be, to timely respond on a nation-wide event/disaster in the most localized way as possible. If I were to also include intellectual property theft on such level, it would be too paranoid to mention, still relevant I think. Abusing the momentum and localizing the attack to target specific users only, would improve its authenticity. For instance, I’ve come across harvested emails for sale segmented not only on cities in the country involved, but on specific industries as well, that could prove invaluable to a malicious attack, given today’s growth in more targeted attacks, compared to mass ones."

The current "events-based" malware is a good example here. If it were a piece of malware to automatically exploit the targeted PC, then you really have a problem to worry about. Meanwhile, Businessweek is running an interesting article on Why Antivirus Technology Is Ineffective, and stating "white-listing" is the future of malware prevention. Could be, if there wasn't ways to bypass the white-listing technology, or give a "white-listed" application a Second Life -- and of course there are.

In another piece of quality research written by Mike Bond and George Danezis, the authors take us through the temptation stage, monitoring, blackmail, voluntary propagation, involuntary propagation, and present nice taxonomies of rewards and blackmail.

And if you're still looking for fancy stats and data to go through, read this surprisingly well written paper by Microsoft - Behavioural Modelling of Social Engineering-Based Malicious Software. They've managed to spot the most popular patterns - generic conversation, non-english language used, virus alert/software patch required, malware found on your computer, no malware found, account information, mail delivery error, physical attraction, accusatory, current events, and free stuff.

Current events, free stuff, and malware on your computer are the most effective ones from my point of view as they all exploit wise psychological tactics. Current events because the Internet is a major news source and has always been, free stuff, due the myth of "free stuff" on the Internet, and the found malware putting the (gullible) end user in a "oops it was my turn to get a nasty virus" state of mind.