00:00:18.720 --> 00:00:23.100
There were 1500 people at the better
Klan. The concert hall was packed.

00:00:24.960 --> 00:00:29.850
I heard a very loud noise at
the back, like firecrackers.

00:00:36.960 --> 00:00:39.690
Alexis LaBrun is one of the
survivors of the bat batter,

00:00:39.690 --> 00:00:44.040
clan terror attack in Paris and
spokesperson of the Life for Paris Victims

00:00:44.040 --> 00:00:44.880
Association.

00:00:49.860 --> 00:00:54.690
So I turned and I saw sparks at
the back and I realized that this

00:00:54.690 --> 00:00:55.523
was gunfire.

00:01:00.420 --> 00:01:05.070
It was the longest hour and a half of
my life. Time was totally distorted.

00:01:21.150 --> 00:01:25.650
There were people dying
and there was crying,

00:01:26.100 --> 00:01:28.650
but as everybody has described,

00:01:29.280 --> 00:01:31.590
there also was some sort of silence.

00:01:33.870 --> 00:01:37.980
The waiting was almost unbearable
because I couldn't see what was going on.

00:01:38.580 --> 00:01:40.380
Someone was covering my view

00:01:44.280 --> 00:01:48.450
when we were finally evacuated.
It was sort of a miracle to me.

00:02:10.650 --> 00:02:14.850
London, Boston, Paris,
San Bernardino, Brussels,

00:02:15.120 --> 00:02:18.450
Orlando niece, WBO Anbar,

00:02:19.020 --> 00:02:23.640
San Re, Berlin, Westminster, Stockholm,

00:02:23.910 --> 00:02:27.180
and Manchester. 14 terrorist incidents.

00:02:27.690 --> 00:02:31.950
32 perpetrators over 400 dead more than

00:02:31.950 --> 00:02:35.010
2,400 injured bombs,

00:02:35.310 --> 00:02:37.590
firearms and ax trucks,

00:02:37.710 --> 00:02:42.360
and a car attackers between
the ages of 17 and 52.

00:02:42.960 --> 00:02:45.660
While the backgrounds of the
attackers are quite diverse,

00:02:45.960 --> 00:02:49.500
there seems to be a pattern
in each attack. At least.

00:02:49.560 --> 00:02:53.760
One of the terrorists was known
to the police in each attack.

00:02:53.940 --> 00:02:58.650
The perpetrators were part
of a network. In each attack,

00:02:58.740 --> 00:03:02.710
the terrorists were
communicating electronically
with each other or with their

00:03:02.710 --> 00:03:05.230
backers. In each attack,

00:03:05.290 --> 00:03:09.610
there was a digital footprint
clearly indicating radicalization.

00:03:10.300 --> 00:03:14.500
How can it be that despite
all these traces and evidence,

00:03:14.980 --> 00:03:19.630
the intelligence services had no
idea that these attacks were being

00:03:19.630 --> 00:03:20.463
plotted?

00:03:22.420 --> 00:03:27.100
What kind of information and what kind
of programs need intelligence services

00:03:27.400 --> 00:03:31.930
to protect us from bombings?
Like in Boston, like Paris,

00:03:31.935 --> 00:03:33.820
like in Brussels,

00:03:35.830 --> 00:03:37.480
mass surveillance is not appropriate.

00:03:39.130 --> 00:03:42.580
Yeah, this is an excellent
question because when we
look at the Boston Marathon

00:03:42.580 --> 00:03:43.810
bombings, for example,

00:03:45.250 --> 00:03:49.150
this was during the height of American
mass surveillance and they were

00:03:49.150 --> 00:03:50.680
collecting the phone calls of everyone.

00:03:50.680 --> 00:03:52.360
They were collecting the
internet activities of

00:03:54.310 --> 00:03:56.680
people indiscriminately just
in case they were useful,

00:03:56.920 --> 00:03:59.800
and it did not stop this attack,
and it did not stop this attack.

00:03:59.830 --> 00:04:04.690
Even though we had been specifically
warned by foreign intelligence

00:04:04.690 --> 00:04:08.560
services that these individuals
were associated with terrorism,

00:04:09.280 --> 00:04:12.670
much the same in Brussels.
So it begs the question,

00:04:12.670 --> 00:04:15.880
why did mass surveillance fail
to catch these individuals?

00:04:16.090 --> 00:04:18.700
It's not because they used encryption.

00:04:18.700 --> 00:04:21.130
It's not because they were
particularly skillful.

00:04:21.340 --> 00:04:23.620
It's not because they were just so clever.

00:04:24.670 --> 00:04:26.980
It's because when you collect everything,

00:04:27.160 --> 00:04:31.120
the communications of everyone in
a nation, you understand nothing.

00:04:31.660 --> 00:04:35.200
You get drowned in so much information
that you can't find what's actually

00:04:35.200 --> 00:04:36.033
relevant.

00:04:36.730 --> 00:04:39.790
Edwards Noden is not the
only one making this claim.

00:04:40.420 --> 00:04:42.880
Among the documents he
leaked to the press,

00:04:42.940 --> 00:04:47.800
there are several internal
memos by analysts at both
the N SS A and its British

00:04:47.800 --> 00:04:49.450
counterpart, G C H Q.

00:04:49.750 --> 00:04:54.580
These analysts complained about the flood
of collection with spot on headlines

00:04:54.970 --> 00:04:59.500
in praise of not knowing too many choices

00:05:01.600 --> 00:05:03.490
and cognitive overflow,

00:05:04.060 --> 00:05:08.860
a tsunami of intercept is declared
with analysts drowning and

00:05:08.860 --> 00:05:10.000
no end in sight.

00:05:16.060 --> 00:05:18.850
I don't fundamentally mind this
sort of mass surveillance approach.

00:05:18.970 --> 00:05:22.390
What I mind is, does it work? Is
it the best way of doing business?

00:05:22.570 --> 00:05:24.820
The problem that we face is
that with mass surveillance,

00:05:24.820 --> 00:05:27.370
we still go and get the intelligence we
want because there's still too much of

00:05:27.370 --> 00:05:31.690
it. So my complaint against
mass surveillance is not
one of civil liberties.

00:05:31.690 --> 00:05:34.900
It's one of efficacy. It just
doesn't work well enough.

00:05:48.400 --> 00:05:53.020
The perpetrator in the Bastille Day
attack in nice Mohamed Boal was a

00:05:53.020 --> 00:05:57.620
petty criminal known to the police
analysis of his mobile phone

00:05:57.770 --> 00:06:01.100
showed that he had contact
with radicals online.

00:06:01.160 --> 00:06:05.630
He declared his approval of the Charlie
UL massacre on the day of the attack.

00:06:06.020 --> 00:06:10.580
He drove his white truck up and
down the promenade despite a ban on

00:06:10.580 --> 00:06:11.720
motorized vehicles,

00:06:12.020 --> 00:06:16.880
despite a heightened state of
alert despite military patrols

00:06:16.880 --> 00:06:18.260
and police controls

00:06:23.360 --> 00:06:25.130
in Wurtzburg Ri Khan,

00:06:25.130 --> 00:06:29.240
AK was in contact with ISIS just
days before his train attack

00:06:29.900 --> 00:06:31.880
in SBA suicide bomber.

00:06:31.880 --> 00:06:36.710
Mohamed Dalil had phone contact with
ISIS Associates minutes before the

00:06:36.710 --> 00:06:37.543
explosion.

00:06:37.610 --> 00:06:41.780
His digital footprint showed a wealth
of radical ideas and statements,

00:06:42.080 --> 00:06:46.970
and he was receiving psychiatric
treatment as a suicide risk in the

00:06:46.975 --> 00:06:50.810
Normandy church attack during which a
priest was killed while conducting a mass,

00:06:51.140 --> 00:06:53.180
both attackers were known to the police.

00:06:53.750 --> 00:06:57.740
Both were in touch with a terror network
on the messaging program. Telegram,

00:06:58.340 --> 00:07:02.270
both were detained in Turkey while
attempting to secretly enter Syria.

00:07:02.840 --> 00:07:06.980
One of the attackers was released from
prison just four months previously

00:07:07.130 --> 00:07:11.990
against the state attorney's explicit
advice and was wearing an electronic

00:07:11.990 --> 00:07:12.823
tag.

00:07:13.130 --> 00:07:17.600
French intelligence received a warning
and a picture of the second attacker

00:07:17.840 --> 00:07:20.060
four days before the attack.

00:07:25.580 --> 00:07:29.000
It's hard to say whether more
data access to more information,

00:07:29.750 --> 00:07:32.900
actually it leads to more
actionable intelligence.

00:07:34.040 --> 00:07:37.970
That's a leap of faith. That's an
assumption that if you have more data,

00:07:38.240 --> 00:07:39.860
then you can make more sense of the data.

00:07:40.280 --> 00:07:43.430
What I would like to see more
of is more targeted efforts,

00:07:43.910 --> 00:07:47.120
more sophisticated strategies
that are not just, well,

00:07:47.150 --> 00:07:48.950
let's have more access to information,

00:07:49.250 --> 00:07:51.680
but let's have better
access to information,

00:07:52.130 --> 00:07:57.020
more targeted information collection
so that we're not letting our small

00:07:57.020 --> 00:07:59.540
teams of analysts drowned in data.

00:08:09.380 --> 00:08:13.520
I was working in nuclear terrorism
for about eight years and before that,

00:08:13.520 --> 00:08:15.770
working in counter-terrorism
for the United Nations.

00:08:19.430 --> 00:08:24.350
Then I started the most popular journal
in the field of terrorism called

00:08:24.350 --> 00:08:29.180
Perspectives on Terrorism
and just gravitated into
the analysis of social media

00:08:29.600 --> 00:08:31.010
and online radicalization.

00:08:44.570 --> 00:08:47.840
I was working with various governments
in Europe on trying to understand

00:08:49.580 --> 00:08:53.120
a new environment that was facing them
when the Islamic State came on the scene.

00:09:03.480 --> 00:09:06.690
The Islamic State is a new
kind of terrorist threat.

00:09:07.530 --> 00:09:11.280
ISIS perfectly masters the global
craving for media attention.

00:09:11.730 --> 00:09:15.570
It acts swiftly and has a
huge presence on social media.

00:09:23.310 --> 00:09:24.540
Set of common.

00:09:24.840 --> 00:09:29.700
The rise of Islamic state in transnational
terrorism brought with it a paradigm

00:09:29.705 --> 00:09:31.740
shift. So on the one hand,

00:09:31.740 --> 00:09:36.240
we have projected terror attacks that
are planned by networks and terror cells.

00:09:36.480 --> 00:09:39.030
On the other hand, we have casual attacks.

00:09:39.060 --> 00:09:43.260
That means radicalized individual
perpetrators carrying out the attack.

00:09:48.240 --> 00:09:53.010
Terrorism is about a lot of people
watching, not about a lot of people dead.

00:09:53.310 --> 00:09:55.740
That's a famous quote from Brian Jenkins.

00:09:56.040 --> 00:10:00.870
Attention is always at the core of what
terrorists want to achieve. Currently,

00:10:00.870 --> 00:10:05.400
their aim is to carry out these attacks
with the simplest possible means.

00:10:28.320 --> 00:10:32.970
ISIS came onto the scene and they were
using social media quite prophetically.

00:10:33.720 --> 00:10:35.370
They were radicalizing individuals online.

00:10:35.370 --> 00:10:36.690
They were recruiting people from Europe.

00:10:37.350 --> 00:10:40.620
I was in the midst of advising
European governments at the time,

00:10:41.310 --> 00:10:44.880
and they came to me and I came to them
with the proposition that we can do more

00:10:44.880 --> 00:10:45.713
on social media.

00:10:53.070 --> 00:10:54.570
From a technical point of view,

00:10:54.840 --> 00:10:58.320
analysis of social media
involves huge networks.

00:10:59.130 --> 00:11:02.310
Robert Wesley first tried
working with existing software,

00:11:02.550 --> 00:11:04.950
but soon discovered that it was too slow.

00:11:05.310 --> 00:11:08.400
His solution was to
develop his own program.

00:11:11.520 --> 00:11:14.070
Started looking for people in 2015,

00:11:15.630 --> 00:11:18.630
so I went out looking for
very specialized individuals,

00:11:19.930 --> 00:11:24.030
mathematicians that also had
hardcore engineering experience,

00:11:25.630 --> 00:11:30.350
artificial intelligence experts
that were very much in turn invested

00:11:30.770 --> 00:11:34.320
in analyzing content
online, analyzing networks,

00:11:34.620 --> 00:11:38.250
and being able to observe behaviors
within networks over time.

00:11:48.270 --> 00:11:50.100
YA is a mathematician,

00:11:50.250 --> 00:11:54.130
systems engineer and software engineer
with a professional background in the

00:11:54.130 --> 00:11:55.300
aviation industry.

00:11:57.580 --> 00:11:59.290
Robert Wesley, who's the C E o.

00:11:59.500 --> 00:12:03.100
Robert Wesley, our c e o,
contacted me via LinkedIn.

00:12:04.630 --> 00:12:06.940
I was looking for
something new at that time

00:12:11.860 --> 00:12:16.750
and he wanted to know my thoughts about
efficient representation and analysis of

00:12:16.750 --> 00:12:18.370
big social networks.

00:12:29.360 --> 00:12:29.620
I.

00:12:29.620 --> 00:12:33.610
Tried immediately to implement my
thoughts mathematically in mathematical

00:12:33.610 --> 00:12:35.590
terminology and while we had lunch,

00:12:35.590 --> 00:12:39.550
I proposed a possible method
and he shook my hand and said,

00:12:39.760 --> 00:12:40.660
welcome on board.

00:12:45.340 --> 00:12:49.660
Christian Wexel Baum is the chief data
scientist at a big data startup in

00:12:49.660 --> 00:12:50.680
Vienna, Austria.

00:12:51.250 --> 00:12:55.120
His background is in software
engineering and artificial intelligence.

00:12:57.670 --> 00:12:57.910
Before.

00:12:57.910 --> 00:13:01.360
Battle Clan, the attacks were more
something that happened to other people.

00:13:01.960 --> 00:13:04.570
There was an attack in
London, an attack in Madrid,

00:13:05.470 --> 00:13:07.240
but it felt as if people
just got used to it.

00:13:10.690 --> 00:13:10.840
When.

00:13:10.840 --> 00:13:12.250
Batter clan happened in France.

00:13:12.550 --> 00:13:16.480
I personally related strongly to the
victims in this case because the attack

00:13:16.480 --> 00:13:19.840
took place at a venue that I could have
gone to as well, an ordinary nightclub.

00:13:21.100 --> 00:13:24.130
I think Batter clan was memorable because
you could really identify with the

00:13:24.130 --> 00:13:24.963
victims there

00:13:26.650 --> 00:13:29.590
and also because as opposed
to London or Madrid.

00:13:30.850 --> 00:13:30.910
You.

00:13:30.910 --> 00:13:34.510
Could follow the batter clan attack
almost as a live stream on social media.

00:13:38.410 --> 00:13:41.770
Batter clan was the first terror attack
that really entered into your personal

00:13:41.770 --> 00:13:46.570
life on your mobile phones, in your
living rooms. You could follow everything.

00:13:46.630 --> 00:13:47.463
As it happened.

00:13:53.950 --> 00:13:58.120
130 Dead 352 injured.

00:13:58.600 --> 00:14:03.220
The Paris attacks in November, 2015 were
the bloodiest attack in France since

00:14:03.220 --> 00:14:04.120
World War ii,

00:14:04.390 --> 00:14:08.740
and they happened despite a heightened
state of alert and mass surveillance.

00:14:09.160 --> 00:14:13.870
The attacks carried out at six
separate locations simultaneously were

00:14:13.870 --> 00:14:18.670
plotted and executed by a well
organized terrace cell linked to isis.

00:14:27.040 --> 00:14:27.873
After the attack,

00:14:28.180 --> 00:14:31.840
I spoke to some colleagues from the area
of artificial intelligence and machine

00:14:31.845 --> 00:14:32.678
learning.

00:14:33.100 --> 00:14:37.060
We wondered if there was anything we
could do because ISIS and the other terror

00:14:37.060 --> 00:14:39.820
organizations were strongly
represented on social media.

00:14:40.810 --> 00:14:41.643
Media.

00:14:42.790 --> 00:14:46.360
So we thought since our strength was
machine learning and natural language

00:14:46.360 --> 00:14:47.193
processing,

00:14:47.290 --> 00:14:50.890
we had some great tools at hand that
could prevent online radicalization.

00:15:02.900 --> 00:15:06.650
At the same time I met Robert.
It was an amazing coincidence.

00:15:07.730 --> 00:15:11.120
Robert came from counter-terrorism and
was looking for someone with highly

00:15:11.120 --> 00:15:12.860
proficient machine learning skills.

00:15:12.950 --> 00:15:13.783
Machine learning.

00:15:13.940 --> 00:15:15.140
We agreed to collaborate.

00:15:16.700 --> 00:15:20.240
Then Robert introduced me to Y and
the three of us met to do our first

00:15:20.240 --> 00:15:21.073
BRAINSTORMINGS.

00:15:21.950 --> 00:15:22.800
Brainstorming.

00:15:26.750 --> 00:15:29.720
How's it going? Hey, how doing, man?

00:15:31.190 --> 00:15:34.760
Our first question was, what kind
of problem do we actually have?

00:15:35.690 --> 00:15:37.880
How can machine learning
help to solve this problem?

00:15:43.490 --> 00:15:43.615
And.

00:15:43.615 --> 00:15:46.520
We discovered that the problem
can be represented in a graph.

00:15:47.120 --> 00:15:51.680
We have actors that radicalize
the connects online that
are active in a social

00:15:51.680 --> 00:15:52.760
network of some kind.

00:16:00.560 --> 00:16:04.580
I can fix him and they're bridged.
Who are they connected to?

00:16:04.640 --> 00:16:08.780
Who's the bridge connected? Yes,
he's connected to the center here.

00:16:08.960 --> 00:16:09.950
Who is he connected to there?

00:16:11.090 --> 00:16:13.310
The world is a network.
Anything is a network.

00:16:16.760 --> 00:16:21.380
Everyone's connected, social media
connects them. Power grids connects us.

00:16:22.310 --> 00:16:24.500
Our personal interactions
offline connect us.

00:16:26.810 --> 00:16:31.370
The ability to represent everything in
life as a network means that we can run

00:16:31.550 --> 00:16:33.200
artificial intelligence. On top of that,

00:16:37.400 --> 00:16:40.940
we can run artificial intelligence to
tell us things that we weren't necessarily

00:16:40.940 --> 00:16:43.220
observing before. That's
what it's all about,

00:16:43.760 --> 00:16:46.520
being able to automate a
lot of the analysis process.

00:16:51.080 --> 00:16:53.180
Machine learning, meet graph theory.

00:16:53.390 --> 00:16:55.490
We combine machine
learning and graph theory,

00:16:55.670 --> 00:16:58.670
but also use genetic algorithms
and probabilistic methods.

00:16:59.090 --> 00:16:59.923
Probabilistic for.

00:17:07.910 --> 00:17:10.430
The data originates in
open source intelligence.

00:17:11.270 --> 00:17:15.110
This is a term that comprises all the
data that's openly publicly available

00:17:15.140 --> 00:17:15.973
online.

00:17:19.160 --> 00:17:19.220
For.

00:17:19.220 --> 00:17:22.460
Example, data used by radical
networks to recruit people.

00:17:26.690 --> 00:17:31.460
ISIS goes online and tries to find
new recruits and to do so they

00:17:31.460 --> 00:17:32.600
spread propaganda.

00:17:33.260 --> 00:17:36.890
They put it online on the web
and it's accessible to anybody

00:17:38.840 --> 00:17:41.240
because their goal is to reach
as many people as possible

00:17:43.000 --> 00:17:44.270
and we're grateful for that.

00:17:44.750 --> 00:17:47.660
We are happy that those people
are so open with their propaganda,

00:17:47.870 --> 00:17:52.470
with their radicalization because
they're basically saying, look,

00:17:52.860 --> 00:17:54.750
this is us and this is our data.

00:17:55.770 --> 00:17:59.520
The fan base of ISIS also provides
great open source intelligence.

00:18:00.390 --> 00:18:00.750
People.

00:18:00.750 --> 00:18:03.390
Openly declare themselves
as supporters of isis,

00:18:04.320 --> 00:18:08.550
connect with other supporters and speak
openly on Twitter about their plans,

00:18:08.580 --> 00:18:10.200
their ideas, their ideology.

00:18:11.460 --> 00:18:11.820
Intelligence.

00:18:11.820 --> 00:18:15.150
This is open source intelligence
for us. This is what we start from.

00:18:26.640 --> 00:18:26.940
This is.

00:18:26.940 --> 00:18:30.900
A perfect method to make assumptions
about radical networks without violating

00:18:30.900 --> 00:18:32.250
the privacy of the users.

00:18:32.670 --> 00:18:34.200
The privacy a user.

00:18:46.980 --> 00:18:49.110
Because everyone knows
that this data is public.

00:18:49.980 --> 00:18:53.160
All of them know when they're writing
on Twitter that Twitter is accessible to

00:18:53.160 --> 00:18:56.880
the public. It's like writing your
support for ISIS onto your house.

00:18:57.210 --> 00:18:58.890
This is what people do on Twitter.

00:19:03.750 --> 00:19:05.730
Named by common websites.

00:19:05.790 --> 00:19:08.460
We also collect data from blogs
and websites on the Darknet.

00:19:12.720 --> 00:19:13.800
Website named Darknet.

00:19:24.120 --> 00:19:24.990
And all this data.

00:19:25.620 --> 00:19:30.210
We can analyze everything without having
to monitor mobile phones or hacking

00:19:30.210 --> 00:19:34.200
emails or computers to check what
people think. This is what we use.

00:19:38.880 --> 00:19:39.713
This.

00:19:40.740 --> 00:19:43.650
This means we're not violating
the privacy of the devices

00:19:45.180 --> 00:19:49.290
and we don't need more
because the information we
can obtain publicly is enough

00:19:49.290 --> 00:19:51.060
to draw extensive conclusions.

00:20:13.680 --> 00:20:18.120
It's amazing how much you can actually
see from publicly available information,

00:20:18.630 --> 00:20:23.190
how much you can predict, how
many suspects you can identify,

00:20:23.970 --> 00:20:27.690
how many individuals you can model the
behavior around who are radicalizing.

00:20:31.330 --> 00:20:31.710
And.

00:20:31.710 --> 00:20:32.543
Years.

00:20:32.760 --> 00:20:36.600
We found several individuals
who are arrested in Europe.

00:20:37.110 --> 00:20:41.400
We had identified them six months in
advance when they stopped their online

00:20:41.400 --> 00:20:42.233
activity.

00:20:52.150 --> 00:20:55.180
Stopping online activity
is a behavioral change,

00:20:55.510 --> 00:20:59.500
and that's a behavioral change that our
software is able to pick up on and we

00:20:59.500 --> 00:21:03.280
can indicate that as an alert to
governments that we don't know what's

00:21:03.285 --> 00:21:06.940
happening. Maybe they've deradicalize,
maybe they've radicalized further.

00:21:07.360 --> 00:21:10.030
Maybe they're going dark and more
secretive with their actions.

00:21:16.390 --> 00:21:19.160
From TEUs.

00:21:20.110 --> 00:21:23.650
We are trying to get a grip on terrorism
with the help of technology and the

00:21:23.650 --> 00:21:26.650
ultimate goal is being able
to predict terror attacks.

00:21:37.190 --> 00:21:40.750
On the 2nd of May, 2016, I wrote
the first lines of the software.

00:21:41.020 --> 00:21:44.290
We were sitting in an empty room At
that time, I didn't even have a server,

00:21:44.350 --> 00:21:47.050
I just had a laptop.
This is where it started.

00:21:48.970 --> 00:21:50.530
The first version of the backend,

00:21:50.770 --> 00:21:54.610
our server-based software was
ready to use at the end of July 20.

00:21:54.610 --> 00:21:59.290
16,016.

00:22:06.040 --> 00:22:08.290
After that, we conducted
our first test runs.

00:22:09.820 --> 00:22:10.065
From.

00:22:10.065 --> 00:22:10.750
This point on,

00:22:10.750 --> 00:22:14.200
we could prove that we would be
able to stick to our design goals.

00:22:14.380 --> 00:22:19.060
That means processing and analysis of
social networks at a high velocity and

00:22:19.060 --> 00:22:23.740
with little reaction time, which
enables us to analyze data in real time.

00:22:26.860 --> 00:22:30.310
I'm talking about milliseconds here.
Whereas others would need minutes or.

00:22:30.340 --> 00:22:31.173
Hours.

00:22:41.150 --> 00:22:44.770
What the team in Vienna was not
aware of at that point in time,

00:22:45.370 --> 00:22:49.000
they weren't the first ones to
have come across that solution.

00:22:49.540 --> 00:22:54.070
A similar system had already been
invented in the labs of the N SS A,

00:22:56.590 --> 00:22:59.170
the mastermind behind it. Bill Binney,

00:22:59.320 --> 00:23:01.690
former technical director of the N SS A,

00:23:03.700 --> 00:23:08.140
but then his program was killed
off shortly before nine 11.

00:23:17.410 --> 00:23:21.700
Welcome to you both and welcome to
the British Parliament, both houses.

00:23:22.030 --> 00:23:26.170
I'm dealing with a very important piece
of legislation that we've been asked to

00:23:26.170 --> 00:23:27.003
look at.

00:23:27.040 --> 00:23:31.990
Do you think actually this bill is
necessary at all and do the provisions

00:23:31.990 --> 00:23:36.640
of the bill strike the right balance
between privacy on the one hand

00:23:37.360 --> 00:23:40.090
and security on the other?
Mr. Binny perhaps to start?

00:23:40.630 --> 00:23:41.110
Yes. Well,

00:23:41.110 --> 00:23:44.650
first I'd like to thank the committee for
giving me the opportunity to come here

00:23:44.650 --> 00:23:48.170
and give some testimony and hopefully
I can help you with some of the issues

00:23:48.170 --> 00:23:51.380
you're discussing here in your
committee. My big objection,

00:23:51.380 --> 00:23:55.850
both with both N S A and
at G C H Q and all of the

00:23:55.855 --> 00:23:59.180
associated law enforcement agencies that
are affiliated with them and how they

00:23:59.180 --> 00:24:03.760
deal with data is fundamentally
circling around bulk acquisition of

00:24:03.830 --> 00:24:05.630
data of any type.

00:24:06.110 --> 00:24:10.400
Why do you think the British government,
with all this G C H Q experience,

00:24:10.460 --> 00:24:13.040
its relationship with the N S A et cetera,

00:24:13.100 --> 00:24:15.680
why has it taken the approach that it has,

00:24:15.740 --> 00:24:19.820
which is so diametrically opposed
to what you are advocating? Why?

00:24:20.270 --> 00:24:22.430
I think I know exactly why, okay.

00:24:22.880 --> 00:24:26.870
They took it because the N S A did and
N S A did it because of contractors and

00:24:26.870 --> 00:24:29.150
the interest in contractors
and getting money and feeding,

00:24:29.270 --> 00:24:32.030
and I accused them publicly.
I've accused them of this,

00:24:32.060 --> 00:24:34.910
of trading the security of the people of
the United States in the free world for

00:24:34.910 --> 00:24:36.980
money and this is why they did it.

00:24:39.110 --> 00:24:43.910
We have heard from a number of law
enforcement officers and security services

00:24:44.150 --> 00:24:49.130
witnesses who are at the rock
face now, not 15 years ago,

00:24:49.580 --> 00:24:53.870
and their evidence has
been that they need these

00:24:53.870 --> 00:24:54.703
powers.

00:24:54.830 --> 00:24:58.790
Are you telling this committee that each
and every one of those witnesses are

00:24:58.790 --> 00:25:01.400
wrong and indeed possibly
misleading the committee?

00:25:01.880 --> 00:25:02.713
I guess I am.

00:25:08.510 --> 00:25:13.310
Bill Benny's evidence leads to Washington
DC A Pentagon report has been written

00:25:13.315 --> 00:25:14.930
about Bill Benny's program,

00:25:15.380 --> 00:25:19.130
but it's still classified 15
years after the investigation.

00:25:19.790 --> 00:25:22.430
Only a few people have
ever read this report.

00:25:22.880 --> 00:25:25.040
One of them is Patrick Edington.

00:25:26.030 --> 00:25:30.620
Edington is a policy analyst in
Homeland Security at the Cato Institute,

00:25:30.860 --> 00:25:34.730
an influential think
tank. In January, 2017,

00:25:35.030 --> 00:25:39.890
Edington sued the US government over
declassification of the Pentagon report

00:25:40.100 --> 00:25:41.900
about Bill Benny's program.

00:25:42.740 --> 00:25:47.300
There are over a hundred
paragraphs in that report that are

00:25:47.305 --> 00:25:51.440
redacted. About 80 of
them are unclassified.

00:25:54.920 --> 00:25:56.510
In spring 2013,

00:25:56.840 --> 00:26:01.340
Patrick Edington was allowed to read the
Pentagon Report and he was immediately

00:26:01.340 --> 00:26:04.130
fascinated by the world
laid out before him.

00:26:05.900 --> 00:26:09.680
I had the chance to read the real
thing, the classified version,

00:26:10.430 --> 00:26:14.270
and it absolutely made my
blood boil the level of waste,

00:26:14.270 --> 00:26:16.100
fraud and abuse that was
exposed in the report.

00:26:16.730 --> 00:26:19.550
The kind of retaliatory measures
that were engaged in by senior N Ss A

00:26:19.550 --> 00:26:21.290
management, all these kinds of things,

00:26:21.950 --> 00:26:24.650
and this is the stuff that they're
basically withholding from the public.

00:26:24.650 --> 00:26:26.900
This is the stuff that they
don't want the public to know.

00:26:29.270 --> 00:26:32.060
The concept behind thin
thread was revolutionary.

00:26:32.480 --> 00:26:35.990
Instead of collecting everything
and then searching for keywords,

00:26:36.380 --> 00:26:40.280
binny used a different approach.
The content of a phone call,

00:26:40.280 --> 00:26:43.040
a text message or an email was irrelevant.

00:26:43.350 --> 00:26:47.370
What was relevant was who
communicated with whom, when,

00:26:47.670 --> 00:26:51.000
how often, and from where. In short,

00:26:51.420 --> 00:26:53.160
the metadata was important.

00:26:53.730 --> 00:26:57.600
The problem was that it was in direct
conflict with another program that the

00:26:57.605 --> 00:27:00.930
agency was pushing downtown with
Congress that was going to cost multiple

00:27:00.930 --> 00:27:01.980
billions of dollars.

00:27:02.310 --> 00:27:07.260
The code name of Thin Thread's
competitor Trailblazer Fin Thread

00:27:07.260 --> 00:27:12.180
was ready for operation in November,
2000. However, N SS a director,

00:27:12.185 --> 00:27:17.100
Michael Hayden and NSA's senior leadership
managed to delay its deployment over

00:27:17.100 --> 00:27:21.600
the course of nine months until it was
finally killed off in August, 2001,

00:27:21.810 --> 00:27:23.940
three weeks prior to nine 11,

00:27:25.200 --> 00:27:26.880
right after the terror attacks,

00:27:27.060 --> 00:27:31.740
the N S A began its mass surveillance
programs and Benny left the

00:27:31.740 --> 00:27:32.573
agency.

00:27:32.760 --> 00:27:37.140
A test run of Thin thread conducted by
Benny's colleague Tom Drake in early

00:27:37.140 --> 00:27:42.000
2002 proved that NSA's
database actually contained

00:27:42.000 --> 00:27:46.650
information about the nine 11 terrorists
and that the N SS A had ditched the

00:27:46.650 --> 00:27:49.980
only search program
able to find that data.

00:27:50.400 --> 00:27:54.480
Conclusion nine 11 could have
been prevented with thin thread

00:27:54.990 --> 00:27:58.140
NSA's response completely
shut the program down.

00:28:05.760 --> 00:28:10.710
It remains one of the most extraordinary
documents that I have seen in my

00:28:10.710 --> 00:28:12.090
almost 30 years in Washington,

00:28:15.090 --> 00:28:18.270
And I've read my share of
investigative reports here.

00:28:18.270 --> 00:28:22.800
You had a small dedicated group of
significant people who came up with a

00:28:22.800 --> 00:28:26.430
completely revolutionary system
for next to no money at all,

00:28:27.660 --> 00:28:31.290
that very likely would've detected the
hijackers before they ever would've had

00:28:31.290 --> 00:28:32.490
an opportunity to strike,

00:28:33.780 --> 00:28:38.100
and they got sandbag by a
bunch of inside the beltway

00:28:38.280 --> 00:28:40.770
bandits with a lot of help.

00:28:41.760 --> 00:28:44.910
This was the worst part from
their own management at N S A,

00:28:45.690 --> 00:28:47.490
including the very senior
leadership of N S A.

00:28:48.180 --> 00:28:51.720
The people who were
responsible for all of this

00:28:53.190 --> 00:28:57.630
thus far at least have gotten off
scot-free. There's been no judgment,

00:28:57.630 --> 00:28:58.800
there's been no accountability,

00:28:59.820 --> 00:29:02.970
and so it simply makes you
wonder at the end of the day,

00:29:03.660 --> 00:29:07.140
what else is going on in N S A and
elsewhere in the intelligence community.

00:29:08.040 --> 00:29:12.720
The British Parliament disregarded Bill
Binney and the evidence he gave before

00:29:12.720 --> 00:29:15.750
the Joint Parliamentary Committee
on the new surveillance law.

00:29:17.640 --> 00:29:19.230
In November, 2016,

00:29:19.470 --> 00:29:24.090
the Investigatory Powers Act was passed
making the UK the most surveilled

00:29:24.090 --> 00:29:25.770
country in the Western world.

00:29:35.850 --> 00:29:36.570
However,

00:29:36.570 --> 00:29:40.950
the Invest Powers Act failed to
prevent both the Westminster attack in

00:29:40.960 --> 00:29:45.400
March, 2017 and the attack in
Manchester two months later.

00:29:47.590 --> 00:29:52.480
Again, the perpetrators in both cases
were already known to the police.

00:29:58.930 --> 00:30:03.220
Several other countries recently passed
new surveillance laws among them,

00:30:03.220 --> 00:30:06.460
Germany, France, Austria, and Switzerland.

00:30:10.630 --> 00:30:14.230
The reasons why governments want to
dramatically expand their surveillance

00:30:14.230 --> 00:30:17.080
capabilities are manifold.

00:30:18.160 --> 00:30:19.630
There's a kind of unholy alliance.

00:30:19.630 --> 00:30:23.500
People who want more spying for one
reason enable people who want more spying

00:30:23.500 --> 00:30:24.130
for another reason.

00:30:24.130 --> 00:30:28.870
So one of the obvious and most banal
reasons is just civil service empire

00:30:28.870 --> 00:30:29.685
building, right?

00:30:29.685 --> 00:30:33.430
If you work in the government and you
can radically expand the mission of your

00:30:33.430 --> 00:30:37.090
department, you get more budget, you get
more stability, you get more reports.

00:30:38.290 --> 00:30:42.880
The org chart focuses on you in a more
meaningful way and you gain more power

00:30:42.880 --> 00:30:44.410
and more long-term power.

00:30:46.780 --> 00:30:47.613
On the other hand,

00:30:47.830 --> 00:30:51.730
technological progress has turned the
world of intelligence agencies upside

00:30:51.730 --> 00:30:52.563
down.

00:30:55.630 --> 00:30:58.180
I was in the British Army for 32 years,

00:30:58.240 --> 00:31:01.690
and over the course of that career,

00:31:02.350 --> 00:31:06.430
the balance of intelligence

00:31:06.430 --> 00:31:07.570
changed.

00:31:08.020 --> 00:31:12.820
It changed from human to sigint

00:31:12.820 --> 00:31:13.660
dramatically.

00:31:14.080 --> 00:31:14.320
Well,

00:31:14.320 --> 00:31:19.120
the image that most people get from
the spy world is of people at the

00:31:19.120 --> 00:31:20.170
C I A who

00:31:24.070 --> 00:31:29.020
go undercover and they find a lot

00:31:29.020 --> 00:31:30.280
of secrets and so forth.

00:31:30.580 --> 00:31:34.780
In reality is spy world
really now is technical.

00:31:34.780 --> 00:31:39.490
It's made up of N S A, which
does a lot of eavesdropping,

00:31:39.520 --> 00:31:41.500
picks up communications around the world,

00:31:41.500 --> 00:31:44.410
is probably the largest
intelligence agency in the world,

00:31:45.280 --> 00:31:49.750
and the other one is the National
Geospatial Intelligence Agency,

00:31:50.590 --> 00:31:54.970
which analyzes all the imagery picked
up around the world and there's

00:31:55.480 --> 00:32:00.220
thousands of miles of film and
billions of pictures that's taken

00:32:00.460 --> 00:32:03.100
by all the drones over the Middle East.

00:32:03.100 --> 00:32:04.960
All the satellites that circle the globe,

00:32:06.550 --> 00:32:09.970
I mean the CI now is
basically running drones.

00:32:20.080 --> 00:32:23.260
But when you dig into how the C
I A directs the drone strikes,

00:32:23.260 --> 00:32:26.410
you find out that they're
basically running another N SS a.

00:32:26.470 --> 00:32:30.310
They do an entire signals intelligence
package to figure out who to kill.

00:32:30.880 --> 00:32:35.500
And it's an interesting question,
why would the C I A sunset?

00:32:35.500 --> 00:32:39.920
Its human intelligence function in favor
of a signals intelligence function when

00:32:39.950 --> 00:32:43.040
obviously governments
need both to do security,

00:32:43.610 --> 00:32:47.690
and I think that thinking
about the private sector
and its incentives is useful

00:32:47.690 --> 00:32:51.950
here because signals intelligence
produces enormous procurements.

00:32:55.850 --> 00:33:00.410
After the US invaded
Iraq, this was 2003, 2004,

00:33:00.860 --> 00:33:04.490
citizens were being dragged in to
be interrogated by US military.

00:33:05.270 --> 00:33:09.110
A lot of the interrogation was
being done by a company called Khaki

00:33:09.110 --> 00:33:09.950
International,

00:33:10.490 --> 00:33:15.080
and it struck me as obscene
that we would have a company

00:33:15.080 --> 00:33:18.680
doing interrogation of
enemy prisoners for profit.

00:33:22.520 --> 00:33:27.170
I looked into this company itself and
learned a lot about the company and who

00:33:27.170 --> 00:33:31.880
ran it and the fact that there was former
high level national security officials

00:33:31.880 --> 00:33:34.010
were helping advise the company,

00:33:34.010 --> 00:33:38.360
and many former military
people were in management

00:33:39.050 --> 00:33:42.860
positions at the company
and they obviously had a
very close relationship with

00:33:42.860 --> 00:33:44.180
the Department of Defense,

00:33:44.720 --> 00:33:49.580
and so this drew me into
investigating why and how so much

00:33:49.580 --> 00:33:51.440
intelligence had been outsourced.

00:33:51.440 --> 00:33:55.970
My best estimate at the
time was that 50% of the

00:33:55.970 --> 00:34:00.230
workforce was contracted, and this
actually seemed pretty high to me.

00:34:01.820 --> 00:34:04.190
A few years later when I
was working on my book,

00:34:04.250 --> 00:34:08.780
I got a document that was an
unclassified document that showed

00:34:09.170 --> 00:34:11.800
the percentage of money going

00:34:13.520 --> 00:34:16.310
into the intelligence community
that was going to private sector,

00:34:16.550 --> 00:34:18.560
and it was actually 70%.

00:34:21.530 --> 00:34:26.030
We have over the last decade been
increasingly concerned about the

00:34:26.240 --> 00:34:28.760
privatization of the federal government.

00:34:29.030 --> 00:34:32.150
This was something that really was
beginning during the Clinton Gore

00:34:32.150 --> 00:34:34.100
administration on a broad scale.

00:34:34.105 --> 00:34:38.570
There was a perception that this would
increase efficiencies in government by

00:34:38.570 --> 00:34:43.040
turning what had previously been work
that was done by federal employees to the

00:34:43.045 --> 00:34:43.878
private sector.

00:34:44.420 --> 00:34:49.220
We have done a lot of analysis that
demonstrates that just on a flat financial

00:34:50.090 --> 00:34:53.810
cost benefit analysis that often it
costs as much as three times as much in

00:34:53.815 --> 00:34:57.230
fact, to be relying on private
contractors than federal employees.

00:34:57.230 --> 00:35:00.350
And when you think about it, it's sort
of reasonable to expect that, right?

00:35:00.350 --> 00:35:03.080
These are private sector businesses
that are trying to make money.

00:35:06.440 --> 00:35:10.940
Mass surveillance is a huge
business in the US alone.

00:35:11.090 --> 00:35:15.860
Surveillance has become a business
that makes 100 billion each year.

00:35:16.610 --> 00:35:18.680
The money comes from secret budgets,

00:35:19.010 --> 00:35:23.630
is handed over to secret contractors
for secret operations and is even exempt

00:35:23.630 --> 00:35:26.180
from legally required
audits of public spending.

00:35:27.260 --> 00:35:32.120
It is an invitation to self-serve
an invitation many are happy to

00:35:32.120 --> 00:35:32.953
accept.

00:35:39.630 --> 00:35:43.920
We have long taken the position that
much of what we see is really legalized

00:35:43.920 --> 00:35:46.620
corruption, so what this means
is it may not be illegal,

00:35:46.710 --> 00:35:48.660
but it doesn't mean that it's not wrong.

00:35:52.770 --> 00:35:53.603
Today,

00:35:53.640 --> 00:35:58.560
five companies control 80% of the entire
private surveillance market in the

00:35:58.560 --> 00:36:01.920
us. Without these corporate
surveillance giants,

00:36:02.280 --> 00:36:06.870
everything would break down at
the N SS A C I A and the 14 other

00:36:06.930 --> 00:36:08.670
US intelligence agencies.

00:36:11.190 --> 00:36:13.920
The private surveillance
monopolies have taken over

00:36:24.450 --> 00:36:28.500
the military. Industrial surveillance
complex is not just an American issue.

00:36:28.890 --> 00:36:31.800
Europe's leading security
researcher, Ben Hayes,

00:36:31.920 --> 00:36:36.570
has outlined how the EU
Commission has since 2003 tried to

00:36:36.570 --> 00:36:41.550
create and foster a European military
industrial surveillance complex in

00:36:41.550 --> 00:36:44.760
order to be competitive on the
international surveillance market.

00:36:53.070 --> 00:36:53.903
For me,

00:36:54.450 --> 00:36:59.340
it all started in 2003 when the

00:36:59.340 --> 00:37:04.200
European Commission appointed
announced that it had established

00:37:04.200 --> 00:37:07.800
out of nowhere what it called
the group of personalities,

00:37:08.700 --> 00:37:13.350
and this this group was set
up to advise the European

00:37:13.350 --> 00:37:15.150
Commission on

00:37:17.100 --> 00:37:21.120
a security research agenda. Now, the
context for this was after nine 11,

00:37:21.960 --> 00:37:26.820
the US government started throwing
money research subsidies to

00:37:26.820 --> 00:37:29.400
big companies to look at
security technologies,

00:37:29.400 --> 00:37:31.950
any kind of technology that
could help keep us safe,

00:37:32.160 --> 00:37:35.940
enhance our homeland security
deal with the threat of terrorism.

00:37:35.940 --> 00:37:40.140
So in the US there was money pouring
into this field for industry,

00:37:40.145 --> 00:37:41.400
for academic research.

00:37:42.120 --> 00:37:45.780
The EU and its wisdom suggested that

00:37:47.970 --> 00:37:52.740
if it didn't essentially replicate
what was happening in the us,

00:37:52.770 --> 00:37:57.120
then we as Europeans
would be at a competitive

00:37:57.120 --> 00:38:00.990
disadvantage because US
companies would be leading.

00:38:00.990 --> 00:38:05.310
What everyone could already see was
going to be a huge lucrative global

00:38:05.310 --> 00:38:08.790
industry, but we would also be at
a security disadvantage because we,

00:38:09.000 --> 00:38:13.530
European states would be forced to buy
technology from Israel, from China,

00:38:13.590 --> 00:38:14.423
from the us.

00:38:18.480 --> 00:38:21.600
In charge of the European
efforts, other usual suspects,

00:38:22.680 --> 00:38:27.330
arms manufacturers such as Les,
E A D S, Saab, fin Mechanica,

00:38:27.540 --> 00:38:29.130
B a E systems deal,

00:38:29.160 --> 00:38:34.110
and Indra plus European telecommunications
companies and IT giants such as

00:38:34.115 --> 00:38:34.948
Erickson, and.

00:38:35.740 --> 00:38:39.850
We need to foster industrial
competitiveness. This
was their key message.

00:38:39.910 --> 00:38:44.260
This is looking at every kind of
element of security, counter-terrorism,

00:38:44.380 --> 00:38:48.430
border control, crisis
management, surveillance.

00:38:55.150 --> 00:38:58.360
The effects of these policies
go way beyond the arms industry.

00:39:01.000 --> 00:39:03.460
The surveillance business is now huge.

00:39:03.730 --> 00:39:06.250
The security business is now huge,

00:39:06.880 --> 00:39:08.950
hundreds of billions of dollars annually

00:39:10.960 --> 00:39:14.500
by border controls, by
air surveillance drones.

00:39:14.770 --> 00:39:19.660
Let us analyze all of this
telecommunications content data that

00:39:19.665 --> 00:39:21.130
you as governments have gathered.

00:39:22.060 --> 00:39:24.940
Let us put C C T V in every street.

00:39:25.240 --> 00:39:28.870
Let us develop satellite surveillance
techniques, whatever it may be. I mean,

00:39:28.900 --> 00:39:30.100
the field is huge.

00:39:30.820 --> 00:39:34.990
You have a position where governments
are kind of desperate to be seen.

00:39:34.990 --> 00:39:37.600
To be doing something,
you've got to be strong.

00:39:37.600 --> 00:39:39.040
You've got to be tough on terrorism.

00:39:39.850 --> 00:39:44.050
The trouble is the public do not know what

00:39:44.260 --> 00:39:48.910
powers the government is handing
over to the police and the

00:39:48.910 --> 00:39:50.170
security services.

00:39:50.710 --> 00:39:55.210
They're asking for things
which they can't even

00:39:55.390 --> 00:39:57.970
explain how it's going to work.

00:39:58.030 --> 00:40:02.080
What is the technology that's going
to be used to siphon up all this

00:40:02.080 --> 00:40:02.913
information?

00:40:03.160 --> 00:40:07.630
Who is going to be in
charge of this system of

00:40:07.900 --> 00:40:11.620
interrogating all these databases?
Because the government don't know.

00:40:11.950 --> 00:40:16.330
The government are asking for a carte
blanche to do whatever they want.

00:40:24.070 --> 00:40:28.720
These recordings were made by an F B
I drone during a Black Lives Matter

00:40:28.720 --> 00:40:30.340
protest in Baltimore.

00:40:35.770 --> 00:40:40.510
There's a whole bunch of extraordinary
research going on since the Arab

00:40:40.510 --> 00:40:45.070
Spring. The Arab spring just came out of
nowhere. It took everyone by surprise.

00:40:47.770 --> 00:40:51.550
You look at the Occupy movement in the
us, which also came out of nowhere,

00:40:51.550 --> 00:40:53.080
alright? It was dealt with pretty quickly,

00:40:53.080 --> 00:40:57.700
but it came out of nowhere and was
suddenly blockading things and shutting

00:40:57.705 --> 00:41:02.260
things down and setting up
huge presences in squares,

00:41:02.260 --> 00:41:06.400
and so suddenly, as a security agency,

00:41:06.910 --> 00:41:09.760
you are saying, well, we
have to try to predict this.

00:41:31.540 --> 00:41:35.090
Many of these things were organized
through social media and spread through

00:41:35.090 --> 00:41:38.930
social media and were grounded in

00:41:39.710 --> 00:41:44.390
internet communication means that
states now have a very strong

00:41:44.395 --> 00:41:45.380
interest in saying, well,

00:41:45.380 --> 00:41:49.940
how can we predict the next
radical domestic movement we have?

00:41:49.940 --> 00:41:53.510
What is it that stops someone who's
going on a demonstration tomorrow from

00:41:54.140 --> 00:41:58.670
setting up the next occupy the day
after? How can we predict this?

00:41:58.670 --> 00:41:59.900
How can we mitigate this?

00:42:10.700 --> 00:42:12.710
Not only Black Lives Matter,

00:42:12.710 --> 00:42:17.420
activists were targeted by state
surveillance on behalf of the

00:42:17.420 --> 00:42:19.040
New York Police Department.

00:42:19.220 --> 00:42:24.140
The Italian company hacking team
hacked the mobile phones of the Occupy

00:42:24.145 --> 00:42:25.250
movement's leaders.

00:42:27.770 --> 00:42:31.760
The Tea party in the US was
subjected to targeted surveillance.

00:42:32.480 --> 00:42:37.370
Anti-war activists became targets,
as did anti-fracking activists

00:42:39.290 --> 00:42:43.610
in the uk. Greenpeace was
surveilled in Austria.

00:42:43.850 --> 00:42:46.910
Animal rights activists
were branded as terrorists.

00:42:50.690 --> 00:42:51.950
Of course, as soon as you do that,

00:42:52.130 --> 00:42:57.050
you are looking at the threats posed
by your own citizens and arguably,

00:42:57.200 --> 00:43:02.030
you are undermining the very democracy
that brings people out on the streets

00:43:02.030 --> 00:43:06.140
and moves them to join movements
like occupying the first place.

00:43:09.440 --> 00:43:14.030
You're taking away the legitimacy of these
things and couching these people as a

00:43:14.030 --> 00:43:18.500
threat to prevent these movements
taking shape or having influence in the

00:43:18.500 --> 00:43:19.333
future.

00:43:31.490 --> 00:43:36.470
If you think that all the rights and
freedoms and wonderful things that we talk

00:43:36.470 --> 00:43:40.850
about, that's where they came from.
They didn't fall out of the sky.

00:43:40.910 --> 00:43:42.680
They came from movements like these,

00:43:42.680 --> 00:43:46.670
from the suffragettes to
police reform, to everything,

00:43:47.090 --> 00:43:49.640
all of the freedoms that
we take for granted now,

00:43:49.645 --> 00:43:53.360
were born out of radical movements
and people doing direct action,

00:43:53.360 --> 00:43:57.830
and if the job of the
state is now to prevent,

00:43:57.830 --> 00:43:58.910
manage,

00:43:59.180 --> 00:44:03.500
nullify these kinds of

00:44:03.500 --> 00:44:08.420
self-organization, then I think
we're in profoundly dangerous times.

00:44:23.030 --> 00:44:25.400
One year after they started
working on their software,

00:44:25.850 --> 00:44:30.200
the team in Vienna has nearly finished
the project. They are running a test,

00:44:31.500 --> 00:44:35.910
an analysis of the ISIS network in
Germany, Austria, and Switzerland.

00:44:49.260 --> 00:44:53.130
If this test with freely
accessible data is successful,

00:44:53.430 --> 00:44:57.360
the software will be ready for use
directly at its source on the web.

00:44:58.410 --> 00:45:03.150
The identities are being anonymized and
represented as combinations of letters

00:45:03.150 --> 00:45:06.570
and digits. After that,

00:45:06.780 --> 00:45:10.590
the actors in the network and their
relationships will be represented in a

00:45:10.590 --> 00:45:11.423
graph.

00:45:18.510 --> 00:45:23.400
The more influential the actor, the
bigger the circle, the more connections,

00:45:23.520 --> 00:45:25.440
the higher the density of the network.

00:45:27.060 --> 00:45:29.190
Individual actors can be marked.

00:45:29.970 --> 00:45:33.390
You can click on them to make their
profile in the social networks appear.

00:45:42.900 --> 00:45:45.990
Also, their direct
connections can be analyzed.

00:45:46.560 --> 00:45:48.960
The profile shows amongst other things.

00:45:49.170 --> 00:45:54.030
This actor's proximity to a network
center as well as the relationship between

00:45:54.060 --> 00:45:54.893
actors.

00:45:57.060 --> 00:46:00.960
The connection between actors can also
be identified in geographic terms.

00:46:01.290 --> 00:46:05.790
As we can see in this example, this
member of a network is based in Raku,

00:46:06.030 --> 00:46:08.850
Syria. The stronghold
of the Islamic state.

00:46:14.340 --> 00:46:18.160
One connection leads to a large
node in the Syrian port of Laia.

00:46:24.870 --> 00:46:25.703
From there,

00:46:25.920 --> 00:46:30.360
there are direct connections to Europe
to another large node in Cologne,

00:46:30.660 --> 00:46:34.170
a small one in Berlin and
another one in Malmo, Sweden.

00:46:38.410 --> 00:46:42.900
There is a connection from Cologne
to another node in the US in

00:46:42.900 --> 00:46:43.733
Philadelphia.

00:46:50.220 --> 00:46:54.390
From Philadelphia, another direct
connection links back to Syria

00:47:00.350 --> 00:47:04.950
from the Syrian node connections lead
to Qatar, Pakistan, and Indonesia,

00:47:10.710 --> 00:47:14.670
the world as a network. That
is what the software depicts.

00:47:24.180 --> 00:47:28.860
The data can also be represented
in a list with just one

00:47:28.870 --> 00:47:32.530
click, you can compile a ranking
of the actors, for example,

00:47:32.740 --> 00:47:34.270
according to their threat potential.

00:47:42.880 --> 00:47:47.500
The team in Vienna is aware that their
program could also be used for something

00:47:47.500 --> 00:47:50.680
other than uncovering networks
of potential terrorists.

00:47:54.340 --> 00:47:55.173
Edward Snowden.

00:47:55.420 --> 00:47:59.800
We know since Ed Snowden that surveillance
tools aren't used only to prevent

00:47:59.800 --> 00:48:03.040
crime, but also for spying
activities of all kinds,

00:48:03.310 --> 00:48:04.600
including spying on your own. Partner.

00:48:08.070 --> 00:48:08.903
Partner.

00:48:12.910 --> 00:48:17.380
The Vienna team's response to the
question of abuse is privacy By design,

00:48:17.680 --> 00:48:21.010
this means that protection of
privacy is built into the system.

00:48:25.620 --> 00:48:26.453
An.

00:48:28.000 --> 00:48:32.710
Antoine de the French author and pilot
wrote about technology and tried to

00:48:32.710 --> 00:48:36.160
answer the question, what
is secure technology?

00:48:36.580 --> 00:48:38.830
What is highly developed technology?

00:48:41.080 --> 00:48:44.740
Since he was a pilot, his answer
focused on aviation technology,

00:48:44.740 --> 00:48:46.600
but it can also be applied. Generally,

00:48:47.050 --> 00:48:51.250
good and secure technology is technology
that I do not have to think about.

00:48:52.210 --> 00:48:54.670
That means if I sit on
a plane as a passenger,

00:48:55.000 --> 00:48:56.740
I don't have to think about the engine.

00:48:59.920 --> 00:49:01.150
Regarding the software.

00:49:01.330 --> 00:49:05.710
This means that metadata analysis
collects only data relating to

00:49:05.715 --> 00:49:06.490
suspects,

00:49:06.490 --> 00:49:11.110
which is encrypted immediately after
a request is made to access the

00:49:11.110 --> 00:49:15.520
data, an independent authority will
generate three keys for decryption,

00:49:15.670 --> 00:49:19.780
one for the police or intelligence
agency, one for the judiciary,

00:49:19.930 --> 00:49:21.910
and one for the legislative oversight.

00:49:22.450 --> 00:49:25.450
Only when all three keys
are employed together,

00:49:25.600 --> 00:49:27.730
can data be inspected and used?

00:49:29.350 --> 00:49:31.360
Every access is logged.

00:49:31.720 --> 00:49:35.380
A digital record is generated
every time data is accessed.

00:49:35.620 --> 00:49:39.730
That way the parties involved have the
chance to justify themselves in a court

00:49:39.730 --> 00:49:42.070
of law. In the event of abuse or error.

00:49:42.370 --> 00:49:47.290
This process is executed in
a few milliseconds and it
is completely automated on

00:49:47.290 --> 00:49:49.810
the basis of precisely defined criteria.

00:49:51.010 --> 00:49:55.960
Never has a program come closer to a
solution that factors in both security

00:49:56.170 --> 00:49:59.890
and the fundamental right
to privacy to equal degrees.

00:50:01.870 --> 00:50:05.290
It depends on how
governments use our software,

00:50:05.710 --> 00:50:10.150
because what our software
does is it allows them to
identify threats very early,

00:50:10.690 --> 00:50:13.390
and then you have a set of options
of how you deal with these threats.

00:50:13.780 --> 00:50:18.460
You send a social worker, do you engage
in the different type of surveillance?

00:50:18.880 --> 00:50:22.960
What do you actually do? Once you've
identified a threat really early in time,

00:50:27.660 --> 00:50:29.690
You can't end terrorism
through technology,

00:50:29.990 --> 00:50:34.010
you can end it with some
sort of easy solution.

00:50:34.370 --> 00:50:36.440
It's a systemic problem for us.

00:50:36.920 --> 00:50:39.410
It's a problem that wasn't
born in the past 10 years.

00:50:39.740 --> 00:50:44.660
There are people within my group of
colleagues who were born well after this

00:50:44.660 --> 00:50:48.440
began. It's existed before
and it'll continue to exist.

00:50:48.740 --> 00:50:51.830
These are societal problems.
They're religious problems,

00:50:52.460 --> 00:50:53.840
they're individual problems.

00:50:54.500 --> 00:50:58.010
People become angry for a variety of
different things and choose a terroristic

00:50:58.015 --> 00:51:01.400
path to achieve their
objectives. It's just a tactic.

00:51:01.670 --> 00:51:04.970
It's a strategy for obtaining
some sort of future vision,

00:51:05.990 --> 00:51:08.990
and as long as it's effective,
it'll continue to be used.

00:51:14.750 --> 00:51:16.790
You want to prevent people
from engaging in terrorism.

00:51:17.960 --> 00:51:19.550
You have to take away
the benefits for them.