20. Memorandum submitted by UKERNA
1. This is UKERNA's written submission to
the Home Affairs Committee's Inquiry into Terrorism Detention
Powers. UKERNA is the not-for-profit company that manages the
operation and development of the JANET computer network connecting
universities, colleges, research establishments and schools to
each other and to the global Internet. This submission considers
only issues relating to the recovery of information from encrypted
computer files and the time required to achieve this.
SUMMARY
2. Four different approaches to obtaining
information from encrypted computer files are considered:
— legal compulsion of the system owner
or other person;
— finding traces of decryption activities
through normal forensic investigation;
— brute-force decryption of the material;
and
— brute-force or intelligence-led attacks
on decryption key passphrases.
If the material was both encrypted and accessed
by people highly skilled in the use of encryption then none of
these approaches appears likely to reveal information without
many years of delay; however if the encryption systems were chosen
or ever used without scrupulous care then it appears likely that
information would be revealed on a similar timescale to a normal
digital forensic investigation.
DETAIL
3. There are at least four different approaches
that might be taken to recover information from encrypted files
on a computer. These are considered in turn.
4. The simplest approach is to require the
owner of the computer, or some other person with access to the
decryption keys, to decrypt the material or provide the key necessary
to do so. Powers to require this, backed by the criminal sanction
of up to two years imprisonment if a person refuses to comply,
are contained in Part III of the Regulation of Investigatory Powers
Act 2000, sections 49 to 56. However, despite the apparent benefits
for investigating authorities, these provisions have never been
brought into force.
5. If the encrypted file has ever been decrypted
on the computer then there is a reasonable likelihood that information
left over from this activity may be found by normal forensic investigation.
This may include clear text versions of part or all of the file,
unprotected versions of decryption keys or passphrases to unlock
those keys and thereby make them available. Many of the routine
processes running on a computer will cause accidental copies of
this and other material to be retained, for example as deleted
files, in filespace related to printing or where the content of
memory has been temporarily saved to disk. Well written encryption
tools will try to reduce the likelihood of this happening, or
to remove such traces when they do occur, however these also require
scrupulous care by the operator to ensure that they do not accidentally
create additional saved information.
6. Probably the hardest method is to attempt
a direct decryption of the material by guessing the cryptographic
key. Using encryption products generally available at present
it is likely to take decades or centuries to blindly guess and
test a significant fraction of the possible keys, and no algorithmic
methods have been published that would significantly reduce this
time.
7. A more productive approach is likely
to be to attempt to defeat the protection applied to the decryption
key. In most encryption systems the key used for decryption is
much too long for a person to remember. This key is therefore
usually stored as a computer file, itself protected by a further
layer of encryption whose key can be remembered. This protection
key is often expressed as a password or passphrase which may be
guessed; the difficulty of doing this will depend on the training,
skill and care of the person who created it and the person who
must remember and use it. Guesses may be based either on computer
algorithms generating large numbers of possible passphrases, or
on information known about the person or found on the computer.
In theory a good passphrase will be as hard to guess as the key
it protects and therefore either approach will still take many
years to have a likelihood of success. In practice the passwords
and passphrases chosen by people are much easier to guess than
this - it is common for more than half of the login passwords
chosen by users of any computer system to be guessed by a computer
program running for an hour or so. Clearly a skilled user should
be able to choose a much better passphrase than this, and these
could take years to defeat.
19 January 2006
| 
