Today, cyber criminals who operate the most sophisticated stealth malware and botnets rely on a remarkably small number of network and hosting service providers, known to the industry as bad actors. These bad actors supply the infrastructure needed to host drive-by download exploits, command-and-control servers, stolen data drop sites, and other more functional network needs such as DNS and reliable uplinks. Having a stable, controllable network allows malware operators to remove one difficult piece of the puzzle and Internet Service Providers (ISPs) are lining up to take their money. Even given that these networks are very well known, it has proven difficult -- in some cases impossible -- to stop cyber criminals and these bad actors due to legal, economic and technical hurdles.
The cyber crime spree that is underway is supported by bad actors that turn a blind eye to the questionable and criminal activities transpiring over their networks. Research from FireEye and others have exposed notorious examples like McColo, ZlKon, HostFresh and many more. The Federal Trade Commission scored a rare victory when it took down 3fn based on findings that 3fn, "recruits, knowingly hosts, and actively participates in the distribution of spam, child pornography, and other harmful electronic content."
However, these bad actors are difficult to bring to justice due to the international nature of their crimes, the slow response time with which they react to shutdowns and the general lack of funding and focus for cyber law enforcement.
Hosting providers in the Eastern Bloc openly market spam e-mail services, ICQ-based spam and spam hosting among their service offerings since they are well outside the jurisdiction of would-be law enforcement. Meanwhile, their U.S.-based equivalents are much more covert, leveraging hosting fronts, multi-national partnerships, IP space sharing agreements and others to hide the real entity behind a business.
Cybersecurity experts say a handful of ISPs and domain name registrars work closely with cyber criminals to set up malicious websites that sell fake software, host and distribute malware, facilitate botnet communications and other important services to perpetrate these online criminal endeavors. Cyber criminals are making billions by holding companies for ransom using DDoS attacks, selling off confidential information, sending phishing spam, as well as selling storage services for pirated movies, music, and illegal images. The monetization possibilities of malware and botnets are so numerous that the creativity of the cyber criminal is the only limit at this point. Underlying all these schemes is a need for a stable cyber infrastructure to provide the criminals with a platform for their various online businesses.
For example, an Estonia company with a very small /24 allocation, Starline Web Services (that is in turn hosted by Compic) was infamous for allowing malicious content on their network. Earlier in 2009, researchers found that 92.62.100.14 was hosting malicious files and drop zones for ZBot, a notorious banking and backdoor Trojan. Also, on 92.62.100.64, they were hosting redirectors used within an iFrame to send victims to exploit sites, such as directlink2.cn (itself hosted on 92.62.100.66) that used a malicious PDF to attack the Adobe Reader plug-in. Notifying upstream providers like Compic about malware they and their customers were hosting usually gave mixed results at best. Complaints were typically addressed only when backed up by some local authorities, say the Estonian Criminal Police. Unfortunately, it is not possible to get law enforcement involved on every abuse complaint and typically only when the problem has already become egregious. In November 2008, the Estonia CERT team directly stepped in to take down a Srizbi C&C hosted on 92.62.100.97 while it was being hijacked by the FireEye research team. It remained down for about four months and popped back up on the exact same hardware and IP in February. This level of arrogance shows the lack of respect that these actors have for their local CERTs.
Another example is ZlKon, hosted by "Datoru Express Serviss, Ltd" and based out of Latvia. They have a single /23 IP address block. Researchers found malware and exploit hosting sites across the entire IP block.
- inetnum: 94.247.2.0 - 94.247.3.255
- netname: ZLKON
- descr: ZlKon
- country: LV
At ZlKon, researchers found the DNSChanger Trojan that overrides ISP settings to reroute traffic through rogue DNS servers and redirectors to take users to exploit sites, fake antivirus sites and other counterfeit software Web sites. On 94.247.2.30 researchers found a site zcounter.cn (reached via a DNS on 94.247.2.38), which then sent malicious javascript that followed up with PDF-based exploits. On 94.247.2.31 (DNS hosted at 94.247.2.40) users were landing on a Rogue/fake antivirus site. On 94.247.2.228 they are hosting a redirector for a Canadian pharmacy scam site. Interestingly, on 94.247.2.183, if the browser's User-Agent string was a Mac, it then served up the Mac version of a fake Flash player -- which again is DNSChanger. The infected Mac then proceeded to phone home to its command and control server on 94.247.2.109. Near the end of the address block, on 94.247.3.251, researchers found patterns consistent with Zbot activities on a domain, innah.cn, where large HTTP POST data transfers were taking place. It's not simply that a datacenter hosts malware -- all datacenters do due to the difficulty of monitoring -- it's that these datacenters purely host malware and no legitimate content.
The technical difficulty in stopping bad actors is also due to their rapid response time to ISP shutdowns or server blacklisting. They have automated their failover processes using malware that is designed to resilient and redundant for high availability. FireEye research into the Srizbi botnet uncovered a failover domain name mechanism that would kick in if the main Srizbi control servers were to be shutdown. Each Srizbi infected computer would connect periodically to a dynamically generated domain name until a new C&C server was contacted. Anyone, be it the original author or downstream "users" with access to the domain name generation algorithm, would then just register the domains in advance of the bots connecting to seek out a new C&C server.
Take the early fight against spam as an example. Today, network service providers, hosting services and security companies rapidly blacklist spam-related IP addresses. Using powerful, but centralized servers to send out spam e-mails is no longer effective. Criminals now send spam via botnets, which make it difficult to create an IP-based spam blacklist. This is because home and corporate systems form the bulk of these botnets, which consist of millions of IP addresses. As such, the trend is towards deeper spam analysis. Almost all spam utilizes malicious URLs to exploit end-users or their systems. Tracking the IP address of servers that host rogue Web sites (e.g., exploit sites, fake pharma sites) has shown to be an effective means of detecting spam by examining the embedded URLs. Other techniques include looking for freshly registered domains, domains that are being fast-flux'd, subdomains of a certain depth, etc.
Criminals are responding by compromising ever larger numbers of systems and hosting malicious content on these home and corporate machines. Essentially, the criminals are rapidly being pushed into "cloud computing" where they decentralize their criminal activities, thus shortening the effective half-life of these malicious server lists. Luckily, this is not happening as quickly as predicted, so naming and shaming bad actors is still an effective means of "cleaning up the Internet."
The third chief obstacle in combating bad actors is the lack of enforcement resources. Abuse desks at legitimate datacenters are by definition a money losing operation, so one could imagine how little effort is put into them at a shadier organization. Hosting providers wishing to maintain a semblance of legitimacy may respond to complaints or pressure from their upstream ISPs to shut down suspected malicious servers. However, the criminals simply pickup and re-establish services elsewhere, either through a sister organization or another less scrupulous hosting provider. More and more we are seeing an international effort among both security researchers and law enforcement, but these efforts are in their infancy.
Decisive action against bad actors has had the most effect when it directly hits their pain point, which as always is their pocket book. Increasing the cost of hosting malware, fake Web sites and C&C servers will go up with the increased risk of prosecution by law enforcement. By raising the transaction cost of doing business, cybercrime can continue to be pushed out of the dark corners of the Internet into more public facing infrastructures. Criminals operate almost totally in the open, which gives no disincentive. If you make things just a little bit harder for them, I believe a sizeable portion of them will reconsider their business model.
This story, "Exposing Bad Actor Sites That Support Cybercrime" was originally published by CSO.