CHAPTER 3 DIGITAL IMAGES (continued)
Requiring the original image to
be produced to the court
3.10 One possibility
for overcoming the problems of digital images would be to require
the production of the original as evidence before the courts (Liberty
p 11). However, as discussed in Chapter 2, a fully electronic
process provides no physical original. Provided the necessary
verification and authentication measures are in place, we think
this suggestion is misplaced. Nonetheless the very suggestion
gives rise to the apparent need to keep original pieces of paper
when documents are imaged for subsequent electronic handling.
3.11 One of the key
implications of the growing availability of digitally captured
and stored images (whether of documents or pictures) is that the
erstwhile reliance on paper and photographic originals (or authenticated
copies of these originals) is no longer always going to be feasible
for use in court. For example the National Automated Fingerprint
Identification System has no original paper source to retrieve
as 'best evidence' in a court of law (p 126). Provided that
verification and authentication measures of digital material are
in place, paper originals will rarely be necessary. Yet the evidence
we have received suggests that there may still be some reluctance
to come to terms with this development. We recommend that the
Government encourage the appropriate legal bodies to draw greater
attention to this change and to widen public awareness that paper
originals are rarely necessary.
3.12 A modification
of the suggestion that only originals should be admissible as
evidence is that where an image has been enhanced or otherwise
altered intentionally, any use of the modified copy as evidence
should be accompanied by the unaltered image (Justice Q 26, Liberty
p 11). The founding testimony which explains the source of
the image would be expected to disclose whether modification had
taken place (eg in the forensic laboratory) and other parties
would be entitled to copies of the unaltered image. But what assurances
are there that the image was not altered before being used as
evidence? There is a danger that legal advisers do not understand
the possible effects of enhancement or they do not recognise that
other evidence for the provenance of the image may be required,
or both. These dangers would be compounded when there are insufficient
resources to secure expert legal and technical advice.
3.13 In this context
proposals were made to us for changes to the law. For example,
that the rules for civil and criminal procedures should be amended
so as to require any person putting forward a digital image as
evidence to disclose to the other side an authenticated copy of
the image prior to processing and details of that process.
3.14 In this, as in
the other areas, we were not convinced that rewriting the law
in an attempt to keep pace with technological change was a sensible
approach. The certainty we have is that technology will continue
to change. No less certain is that the rate of change will increase.
Thus, we agree with the Data Protection Registrar (Q 222)
that attempting to extend legislation by introducing specific
technological caveats to take account of the latest technology
is unlikely ever to be a successful approach.
Excluding the
use of digital images as evidence unless specific requirements
are met
3.15 Some witnesses
(eg Liberty p 4, Mr Sommer, p 45) suggested that digital
images should only be admitted as evidence if the technology used
to capture those images incorporated some form of authentication
technology such as watermarking or a digital signature. Indeed,
Mr Sommer went as far as saying that "without authentication
digital images are worthless".
3.16 We have outlined
earlier the existing requirements for admissibility of evidence
(see Chapter 2). The arguments against specifying new criteria
which must be met before evidence can be admitted are:-
it would be very difficult to specify
the nature of the authentication technology in such a way that
it would not quickly become outdated as the technology advances.
it would take an appreciable time
for manufacturers of digital image technology to incorporate such
measures, and even longer for such technology to become widely
used;
when technology advances, the courts
will be faced with the position that images over which there was
no dispute as to their reliability cannot be received as evidence
because they were not captured by technology which met the required
specification; and
the clear trend in the development
of the law is to remove prior requirements for all forms of documentary
evidence, leaving it to the courts to determine whether the evidence
is in fact reliable.
3.17 For these reasons
we are not convinced that some sort of criteria must be met before
evidence can be admitted. Rather we agreed with the witnesses
who said that there should not be different rules about admissibility
based on the technology used to capture the evidence (eg IBM Q 306).
There might be occasions where the most mundane home video could
produce very useful evidence for the court which would be unlikely
to be disputed. Neither would we wish to exclude the vast majority
of business documents and other images in use today which may
not meet some technical standard or other (cf Mr Smith Q 95, IBM
p 99). Similarly, one can hardly expect those engaged in
criminal activities to conform (Mr Sommer Q 118)!
3.18 We recommend
that evidence should not necessarily be inadmissible because it
does not conform with some specific technological requirement.
3.19 But, although
we recommend that there should be no technological requirements
which all digital images must meet before they can be admitted
in evidence, this does not mean we are against authentication
technologies. It is a matter for the courts to determine the evidential
weight and reliability of an image, including the effects of any
modification. We support the application of any technology which
can help with the verification of an image and provide assistance
to the courts in assessing the worth of the evidence before it.
3.20 We recommend
that the Government encourage the use of authentication techniques.
Members of the legal profession should be made aware of the benefits
of these techniques, their value in adding weight to evidence
and the possible significance of their omission.
3.21 Technical procedures
are only part of the authentication process or audit trail. In
addition the provenance of evidence can be enhanced by procedural
measures. Good practice for handling digital documents has been
set down by the British Standards Institution[22]
but conformance with this standard is perhaps not as widespread
as it could be. Certainly it is good business practice to conform
to such standards in preparing documents that might be used in
court. We recommend that the Government produce guidance on
the benefits of conformance with procedural measures necessary
to establish the reliability of evidence, with particular reference
to existing standards. When this guidance is available, we recommend
that the trade associations of those organisations likely to be
concerned with it produce training material on its use.
Identification
in the Criminal Court
3.22 A real danger
posed by digital image technology is that of misidentification,
particularly in criminal prosecutions. The law is already alive
to this danger and contains safeguards to ensure that undue weight
is not given to identification evidence. In the 1960s a series
of miscarriages of justice were caused by mistaken identification
of defendants, and subsequently in the case of R v. Turnbull[23]
the Court of Appeal (a full Court) laid down guidelines to be
observed by judges when 'identity' is an issue. This has become
known as the Turnbull Warning.
"First, whenever the case
against an accused depends wholly or substantially on the correctness
of one or more identifications of the accused which the defence
alleges to be mistaken, the judge should warn the jury of the
special need for caution before convicting the accused in reliance
on the correctness of the identification or identifications. In
addition, he should instruct them as to the reason for the need
for such a warning and should make some reference to the possibility
that a mistaken witness can be a convincing one and that a number
of such witnesses can all be mistaken ...
Secondly, the judge should direct
the jury to examine closely the circumstances in which the identification
by each witness came to be made. How long did the witness have
the accused under observation? At what distance? In what light?
Was the observation impeded in any way, as for example, by passing
traffic or a press of people? Had the witness ever seen the accused
before? How often? If only occasionally, had he any special reason
for remembering the accused? How long elapsed between the original
observation and the subsequent identification to the police? Was
there any material discrepancy between the description of the
accused given to the police by the witness when first seen by
them and his actual appearance? If in any case, whether it is
being dealt with summarily or on indictment, the prosecution have
reason to believe that there is such a material discrepancy they
should supply the accused or his legal advisers with particulars
of the description the police were first given ... Finally [the
judge] should remind the jury of any specific weaknesses which
had appeared in the identification evidence ...
When, in the judgement of the trial
judge, the quality of the identifying evidence is poor ... [he]
should then withdraw the case from the jury and direct an acquittal
unless there is other evidence which goes to support the correctness
of the identification ..."
3.23 More recently,
the Court of Appeal has recognised that similar dangers are posed
by video evidence of identity:
"some at least of the considerations
underlying the safeguards built into the regulatory procedures
laid down for identity parades and the showing of photographs
come into play too with regard to the showing of video tapes.
Particularly in cases where the quality of the video is poor or
the opportunity it provides for recognition is limited ... it
would be desirable to regulate its showing so as to maximise the
prospects of any recognition evidence being truly spontaneous
and independent and minimise the risk of anything being said or
done which might infect that independence and spontaneity and
instead prompt the recognition of some particular person ... It
is not for this court to suggest what the appropriate procedures
should be; clearly regard will need to be had to the practicalities.
We would, however, urge that some immediate thought be given to
this matter at a high level"[24].
3.24 In 1995, R
v. Clare and Peach[25]
demonstrated the ability of the law to adapt to video technology.
The defendants were charged with offences of violence at a football
match, the incident having been captured on an unclear, black-and-white
video recording. A police officer compared this recording, frame
by frame, with a high-quality, colour recording made of the crowd
entering the stadium, and was thereby able to identify the defendants
as having taken part in the violence. The Court of Appeal held
that their conviction on this evidence was safe because an appropriate
version of the Turnbull Warning was given to the jury. Lord Taylor
said:
"... as technology develops,
evidential practice will need to be evolved to accommodate it.
Whilst the Courts must be vigilant to ensure that no unfairness
results, they should not block steps which enable the jury to
gain full assistance from the technology"[26].
3.25 To date the courts
have not developed a definitive statement of the safeguards which
will be necessary where video evidence has been collected by a
source other than a public body, or where image enhancement has
been attempted as part of the identification process.
3.26 A further consideration
arises in those cases which may be equally dependent on identification
but where there is an image that unambiguously shows the defendant.
It is not a matter of caution whether there can be a mistake in
identifying the person whose image is submitted as evidence, of
that there might be no doubt: it is a matter of caution over the
probity of that evidence.
3.27 We recommend
that consideration be given to the Turnbull Warning being appropriately
adapted so that the uncertainties inherent in images as evidence
are made clear to the jury, particularly the implications of any
measures to substantiate authenticity and breaks in the audit
trail, and any processing which the image has undergone. This
is a matter which the Judicial Studies Board may care to take
up.
The current need
for greater certainty
3.28 It can be argued
that existing law is likely always to be adequate and, where there
is any uncertainty, this can be resolved by the courts in the
normal process of building up case law. But allowing the precedents
of case law to establish the legal requirements, is a lengthy
and for the trailblazing litigant an expensive process. We noted
a marked reluctance of both civil (Abbey National, QQ 324, 344)
and criminal authorities (City of London Police) to be the first
to become involved in litigation about digital images. If no cases
are heard, no case law is developed-and uncertainty grows. With
large investments hinging on the outcome, case law is not an ideal
way of establishing the necessary legal confidence. A number of
witnesses, including the Abbey National (QQ 375-379), the BSI
(QQ 196-200) and Mr Sommer (Q 117) called for a code of practice
or guidance for handling computer derived evidence: something
"to give the managers confidence to take that leap and start
to move away from this paper culture that we are now in"
(Abbey National Q 376). The uncertainty must be resolved. The
matter could be tested by a claim for a declaration in the High
Court by means of a friendly action brought by a major institution;
alternatively it could be the subject matter of consideration
by the Commercial Court Users Committee; or the Government might
consider issuing guidance, which of course would not be binding
on the courts. We recommend that consideration be given to
these matters and the Government ensures there is a satisfactory
resolution of the uncertainty.
Maintaining confidence
in the accuracy of the image
3.29 A danger frequently
pointed out to the Committee was that of portraying image enhancement
as an entirely objective science, particularly where an operator
was manipulating the image to look for a suspect. It can become
a largely subjective exercise. It might be possible, with certain
software (eg the Home Office 'IMPROVE' software), to reproduce
any changes, but this makes them no less subjective. Put at its
simplest: if an operator is looking for a white car there is a
temptation to increase the brightness of the image of a grey one;
conversely when looking for a black car, the brightness is reduced
(Liberty Q 5). This might occur without any malice. It is a simple
desire to find what is being searched for: "the well intentioned
mistake" that could lead to a miscarriage of justice (Mr
Sommer Q 123). However we were told by the Home Office (p 127)
that although the use of IMPROVE had been widespread since 1990,
it had not been challenged as evidence.
3.30 Where the record
has been altered, or the technology allows the possibility of
alteration, the user will need to be able to prove that the image
is authentic. Doubts will arise where there is a possibility that
through human or machine error the record has become corrupted
or confused with some other record[27].
Doubts will also be raised if fraud is suspected but, and this
is an important point at the centre of many of our recommendations
on awareness, the apparent credibility of an altered digital image
may mean that fraud is not considered in the first place (eg Mr
Sommer Q 140).
3.31 Furthermore, there
is a danger that an altered image could be given an authenticating
mark if the authentication were not done at the time of image
capture. The value of such technologies would then be compromised.
We also recognise that any authentication technology, however
sophisticated, has the potential for being circumvented. Authenticating
technologies will thus need to keep ahead of the abilities of
those who set out to undo them. But these apparent shortcomings
are possibly less real in practice than in theory. Our witness
from IBM said of circumvention technologies: "in all of these
processes the only thing that you can do is make it very difficult,
and if you can make it difficult enough, such as that the process
takes too long, then you are at least achieving part of your aim"
(Q 292). We concluded that watermarks can be used to great advantage:
they can provide a high level of security in conjunction with
an audit trail and the cost of introducing a watermark to an image
is likely to be low relative to the costs of trying to circumvent
it. Of course the veracity of the witness presenting evidence
should be considered alongside the evidence itself (QQ 99, 118).
3.32 What is important
is to encourage the adoption of authentication technologies by
creating a wide understanding of the benefits from the level of
security they can provide. The technologies are advancing rapidly
and there will also be the need to provide regular updates to
both the users and suppliers on improvements and changes. The
Government already gives 'type approval' for certain enforcement
technologies using images as evidence and we see the measures
set out above as an extension of this.
3.33 We recommend
that the Government encourages the adoption of technological measures
for the authentication of images as evidence by giving type approval
to them. The Forensic Science Service should provide ongoing advice
for manufacturers and users of imaging equipment on authentication
technologies.
3.34 We noted in paragraph
3.6 that hitherto audit trails have not been demanded in cases
where video images were used by the prosecution and in paragraph
3.11 we have made a recommendation on the need for greater public
awareness of the particular circumstances relating to the need
for paper originals. Overall we see a need for awareness by all
those concerned with the potential difficulties with digital evidence.
We recommend that the Judicial Studies Board consider establishing
a programme of education on the implications of digital technology
for the judicial system.
3.35 But if there is
increased awareness, the defence may have insufficient financial
resources to undertake technical investigations and to secure
expert witnesses capable of challenging the reliability of an
image. This is likely to be a particular problem where the defence
is funded by legal aid. What is needed is some mechanism to make
compliance with good practice, conformance with codes of management
and the application of technological measures, accepted normal
practice.
3.36 Voluntary standards
and codes exist for the procedural measures necessary for document
handling (eg the BSI code). We have recommended that similar codes
and standards be put in place to encourage the use of technological
measures to authenticate images. The advantage of such voluntary
standards is that they can be changed to keep pace with changes
in technology (or made substantially independent of it), they
represent the consensus of those in the field and enable a far
more flexible approach than is possible with regulation. But to
achieve the full benefits from such codes there must be an incentive
to comply with them. This is achieved in other areas by making
compliance with an approved code a recognised way of showing that
legal requirements have been met or that good practice has been
achieved. We recommend that the Government devise incentives
or put in place measures such as the endorsement of relevant codes,
to increase the adoption of good practice.
22 BSI DISC PD 0008, see footnote to paragraph 3.2. Back
23
R v. Turnbull (1977) QB 224 Back
24
R v. Caldwell & Dixon
(1993) 99 Cr App. Rep. 73, 78 per Brown Wilkinson LJ. Back
25
R v. Clare and Peach (1995)
2 Cr App. Rep. 333. Back
26
(1995) 2 Cr App. Rep. 333, 339. Back
27
Such doubts can be overcome by producing evidence as to the technical
efficiency of the system, the maintenance of logs sufficient to
indicate problems, error-checking procedures, management and training
of staff, etc. Adherence to technical standards will assist in
authentication, although even standards which are designed to
achieve authentication cannot be guaranteed to do so, as English
law has no system of certification which would raise a presumption
of authenticity. Back
|