For MI5, it has contributed significantly to the disruption of terrorist operations
and to the saving of lives. GCHQ gains benefit from merging the data with
the product of bulk interception, and claims to use the power across the
range of its operational work, though we saw only one case study to illustrate
this. The power is useful in eliminating lines of enquiry and so focusing
resources where they are needed. It is extensively used on a daily basis,
and contributes material to some 5% of GCHQ’s intelligence reporting.252
(c) The bulk EI power is not currently authorised and has never been used,
though targeted EI is seen as an important capability across the full range of
threats to the UK, driven by increasing use of encryption and diversity of
communications methods. EI already contributes to some 20% of GCHQ’s
intelligence reports, and more than a third of higher grade intelligence. The
thematic EI power provided for in Part 5 of the Bill (which is subject to fewer
limitations than the proposed bulk power) has been used at scale to identify
dangerous extremists in Syria. Bulk EI is likely to be only sparingly used,
and (like thematic EI) will require particularly rigorous and technicallyinformed oversight. But I have concluded that there is a distinct (if not yet
proven) operational case for bulk EI in relation to counter-terrorism, counterproliferation and cyber-defence.253
(d) BPDs are used on a daily basis, particularly by MI5 and MI6 where internal
documents show that they are viewed as “critical”, “essential” and
“fundamental”. We were shown their utility in identifying possible MI6 agents,
hostile state actors and potential terrorists, including individuals who posed a
threat to the London Olympics and to the UK in the wake of recent attacks in
France and Belgium. We also observed how they can be used to exclude
large numbers of people from an investigation and to enrich information
obtained by other means. The operational case for them is evident.
(e) While alternative capabilities could sometimes be deployed, including
targeted versions of the powers under review and the use of human agents,
they were likely to produce less comprehensive intelligence and were often
more dangerous (for example to agents and their handlers), more resourceintensive, more intrusive or – crucially – slower. In many cases, there was
simply no realistic alternative to use of the bulk power. I concluded that in
the great majority of the case studies to which we were introduced, the
252
253
See chapter 6 above, in particular 6.9-6.11 and 6.44-6.49.
See chapter 7 above, in particular 7.10, 7.23-7.26 and 7.32-7.38.
123