text
stringlengths 4
429
|
---|
Incursion phase
|
The actual break-in occurs during this phase. The attacker usually compromises the network by delivering targeted
|
malware to vulnerable systems or employees. There are two main avenues of attack. One is to send spear phishing
|
emails, where a link to a malicious website or a malicious attachment is delivered using social engineering
|
techniques. The second method, which is gaining traction, is watering hole attacks, where the attacker infects a
|
website that has a high likelihood of being visited by the intended victim. By using IP address filters before infecting
|
any visitor of such sites, the attacker can reduce the number of infected systems and bring it to a manageable
|
quantity which can be assessed manually at another time.
|
Some groups carefully plan watering hole attacks. For example the Hidden Lynx group stopped using a zero-day
|
vulnerability in a large watering hole attack after Microsoft released details on the vulnerability. This helped to cover
|
their activities and avoid unwanted attention. A few days later the group resumed the watering hole attack again,
|
this time using a different exploit.
|
For more difficult targets, man-in-the-middle attacks can be used. These can be performed either at the same
|
physical location, posing as a genuine Wi-Fi hotspot or through supply chain attacks. This can enable the attacker to
|
swap an update of legitimate software for a maliciously crafted version. Once the victim installs the genuine looking
|
update, the attacker effectively gains control over the computer. Due to the complexity of such an attack, they are
|
rarely used. Depending on the skills of the attacker and the time available, the attacker might also attack systems at
|
the perimeter, such as Web servers, and try to break in from there.
|
The malware used is not always sophisticated. Sometimes a regular off-the-shelf back door Trojan is used. In
|
these cases the person behind the malware orchestrating the commands is what makes the difference between a
|
targeted attack and a broad generic infection. Having said this, on very unique targets, we will often see the use
|
of a specifically designed piece malware, such as in the case of Stuxnet. Depending on the protection measures
|
implemented by the target, the attackers may also digitally sign their malware creation. In the past there have been
|
quite a few cases where code signing certificates were stolen and later misused to sign malware in order to pass it
|
unnoticed to high value targets.
|
Discovery phase
|
Once the attacker has a foothold on one system, the next step is to create a plan for lateral movement through the
|
network until the interesting data is found. With more specialized teams of attackers, we can often observe that the
|
infected system is first analyzed to ensure that it is of interest to them. With watering hole attacks especially, it can
|
happen that computers that were not targeted get infected. Infected computers need to be assessed by the attacker
|
and, if necessary, removed to keep the profile, and with that the chances of exposure, low.
|
One of the obvious tasks performed by attackers is to install key loggers, dump local credentials, search local
|
storage for saved accounts and sniff the network for passwords. Any account detail can be useful to them. Domain
|
administrator passwords are of especially high value, as they can help greatly in moving further through the Intranet.
|
Often small scripts or even manual commands are used to comb through local files and create network mappings.
|
Simple system commands can help the attacker to learn about installed security tools, saved links to internal
|
Page 26
|
Targeted Attacks Against the Energy Sector
|
platforms and local address books.
|
Once new systems are identified the
|
attacker will attempt to hop onto
|
them as well. In some instances
|
they might even use zero-day
|
vulnerabilities to spread further into
|
the network.
|
One method which is gaining more
|
relevance is the hijacking of local
|
software distribution systems for
|
further distribution. This can either
|
be proprietary systems, such as the
|
case of Trojan.Jokra in South Korea,
|
or OS-specific, such as hijacking
|
Windows Update, in the case of
|
Flamer. Once the attackers have
|
successfully managed to create
|
and distribute their own package, Figure 9: Typical commands used during discovery phase
|
they can easily infect all connected
|
systems at once. Especially in cases
|
of wiping attacks, such Trojan.Jokra, this is a very efficient way to disrupt as many computers as possible.
|
If the target is assumed to be in a separated network not connected to the Internet, the malware used might try and
|
autonomously infect removable drives, like USB sticks, or project files for PLCs. This could allow the malware to be
|
manually introduced to the destination network, without the knowledge of the carrier, essentially jumping air gaps
|
into isolated networks.
|
At the end of the discovery phase the attackers should know the internals of the infected networks and have
|
identified systems with interesting data or with connected industrial control systems.
|
Capture/exfiltration phase
|
The capture and exfiltration phases are not always present. If the sole goal of the attackers is to cause a disruption
|
they may directly jump to a destructive payload. However, in most cases information is extracted first, which in turn
|
allows the sabotage to be constructed more efficiently at a later phase.
|
In this phase the interesting data is gathered and sent back to the attackers. This can be done with different levels of
|
sophistication. The simple attacks compress the files and upload them through FTP or through a HTTP POST request
|
to a remote drop server. More sophisticated attackers obfuscate the data by XOR-ing it, encrypting it with proper
|
asymmetric encryption or embedding it into media files using steganography to hide the data from traffic inspection.
|
In addition to this, the amount of data sent and the timing can be chosen in a smart way. For example, some malware
|
samples will send the data in smaller bursts so as not to swamp the network or generate network spikes that might
|
attract attention. Since most employees use laptops, the malware can use location awareness to detect if the
|
compromised computer is outside of the corporate network and send the data once it
|
s directly connected to the
|
Internet, such as from a Wi-Fi hotspot at an airport. This might allow the traffic to bypass perimeter security and
|
receive less scrutiny. In some instances the infected computer might not have a direct connection to the Internet. In
|
such cases, a previously compromised computer in the DMZ can act as a proxy, forwarding all the collected data.
|
Disruption phase
|
This is when any destructive payload is launched. If the attackers are only after information this phase might not
|
happen at all. The targets and the goals for disruption attacks can be very different, there is no such thing as one-
|
size-fits-all for disruption attacks. For example, Stuxnet was tailored to attack a specific uranium enrichment facility
|
and would not work against a different target.
|
In recent times, wiper Trojans have been popular in attacks against the energy sector. The malware deletes all files
|
Page 27
|
Targeted Attacks Against the Energy Sector
|
on a computer and then deletes the master boot record, rendering the computer unusable. This can happen on
|
any operating systems and we have seen scripts for different UNIX flavors being used as well. Depending on
|
the disaster recovery plan in place, these computers can be remotely recovered. However, there may still be an
|
outage while the computers are being restored.
|
Resources
|
[1] Marco Cova, Corrado Leita, Olivier Thonnard, Angelos D. Keromytis, and Marc Dacier. An analysis of rogue
|
AV campaigns. In Proc. of the 13th International Conference on Recent Advances in Intrusion Detection
|
(RAID), 2010.
|
[2] O.Thonnard, M.Dacier. A Strategic Analysis of Spam Botnets Operations. CEAS
|
11, Perth, WA, Australia,
|
Sep 2011.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.