Chapter 2/ Overview of Anti-Forensics
Data Fabrication
Data fabrication is a truly devious strategy. Its goal is to flood the forensic
analyst with false positives and bogus leads so that he ends up spending most
of his time chasing his tail. You essentially create a huge mess and let the
forensic analysts clean it up. For example, if a forcnsic analyst is going to try
and identify an intruder using file checksums, then simply alter as many files
on the volume as possible.
Data Source Elimination
Sometimes prevention is the best cure. This is particularly true when it comes
to AF. Rather than put all sorts of energy into covering up a trace after it gets
generated, the most effective route is one that never generates the evidence to
begin with. In my opinion, this is where the bleeding edge developments in
AF are occurring.
Rootkits that are autonomous and rely very little (or perhaps even not at all)
on the targeted system are using the strategy of data source elimination. They
remain hidden not because they've altered underlying system data structures,
as in the case of active concealment, but because they don't touch them at
all. They aren't registered for execution by the kernel, and they don't have a
formal interface to the I/O subsystem. Such rootkits are said to be stealthy by
design.
2.4 General Advice for AF Techniques
Use Custom Tools
Vulnerability frameworks like Metasploit and application packers like UPX
are no doubt impressive tools. However, because they are publicly available
and have such a large following, they've been analyzed to death, resulting in
the identification of well-known signatures and the construction of special-
purpose forensic tools. For example, at Black Hat USA 2009, Peter Silberman
and Steve Davis led a talk called "Metasploit Autopsy: Reconstructing the
Crime Scene"'-� that demonstrated how you could parse through memory and
recover the command history of a Meterpreter session.
13. http://www.blackhat.coni/presentation!i/bh-usa-09/SILBERMAN/BHUSA09-Silberman-
MetasploitAutopsy-PAPER.pdf.
48 I Part I