Tip:
Highlight text to annotate it
X
Welcome to the session today on how TASER is
working to help with Microsoft to help build a cloud for
law enforcement purposes.
My name is Paul Rosenzweig.
I am a senior advisor in the Chertoff Group, which is
a cybersecurity consultancy up in Washington, DC.
And with me on the panel today, I'm the moderator, with me on
the panel today immediately to my left is Stuart Mackey who is
the U.S. National Technology Officer for Microsoft.
Responsible for shaping and articulating
Microsoft's technology, vision and strategy especially
with relationship to outreach state and local law enforcement.
He served as Chief Informational Officer of Washington state,
which is the original digital state, and
he was also the Vice President of Walt Disney.
And to the far left is Jenner not Jennifer.
Jenner Holden who has ten years experience in evaluating and
developing and managing enterprise level information
systems and right now, he is at TASER International,
where he is responsible in large part for the development of many
of the programs we're gonna be talking about today.
So we're here for about 45 minutes till 3:30,
we will have little bit of time for Q&A at the end.
And we're not going to do this speech-wise, we are going to do
it conversational style as a way of kind of bringing you into
the discussion that has occurred between Microsoft and
TASER about some of the cloud based developments that they've
been doing.
So I want to start with you Stuart and
just ask you to begin.
Give us a little bit of an overview.
And tell us about how Microsoft's long-term commitment
to law enforcement has developed where we are today.
What is kind of the arc of Microsoft's working with the law
enforcement community that brought you to this place
right now?
>> Okay, thanks, Paul.
First of all,
I want to say to everybody's expectation that nobody's gonna
get tased on stage today.
Although, we did talk about it as a possibility.
And we'd certainly be able to pack a room if-
>> We can discuss this Stuart
later.
>> If you're bad, you might get tased.
>> Yeah, well I'm gonna be really careful up here.
In all seriousness too, I wanna thank everybody for
taking time to come here.
I know it's a significant committment of time and
energy for people to come to events like this,
and I know, get a chance to talk to a couple people,
appreciate San Bernardino being here as well,
kind of a surprise to run into you, and
I'm looking forward to talking to you some more.
Let me kind of set the stage, actually, and
even maybe go a little broader than law enforcement in that
over the last 30 plus years Microsoft really has a long
history of working with enterprise customers at all
levels including government and
dealing with some of the most difficult or challenging
data requirements, regulatory environments and such.
And really that 30 years represents kind of this
evolution into the digital world.
And what we mean by that is you know from the evolution
of the PC to the connected world where you know we have super
computers in our pocket connected to everybody else
realtime.
It's something that we've been involved with all along and
very, very much committed to helping enterprise customers and
government in general.
One of the things that we recognized a few years ago was
really with the digitization of information from all kinds of
things, that it was require a new level for us to commit to,
really in this networked world, in this cloud services world.
I really like the word hyperscale.
Cloud computing as kind of this ambiguous fluffy thing,
everybody got a cloud it is.
Nobody really knows what it means but
for the context of Microsoft it's really about massive
computing infrastructure.
Massive storage infrastructure and our history in the past has
really been about protecting information, protecting data,
providing secure systems from the front end.
Whether it's Office and applications and
Word and Email that see you on your desktop to the middle tier,
technologies like dynamics building solutions.
To the back end which was Windows Server and now we
call Azure in the cloud, and so we have really a long history of
addressing some really critical compliance issues.
What we've done related to law enforcement is we really
recognized a high watermark, and
that as we went to look at various compliance capabilities,
our challenges that were out there standards.
We actually identified the law enforcement standard or
CJIS which is driven by the FBI CJIS division through a policy
board called the APB.
That was really a fairly significant in high water mark.
And the idea for us to address that was to, if we address
the high water mark then the Blueberry Commission and other
less stringent requirements would have the ability and
feel confident that if that kind of data works then that's okay.
The one thing that's really happened that's been incredibly
interesting.
I would like to say we were very brilliant and
anticipated all this.
And we clearly had prepared.
But it's bigger than we even imagined.
Was frankly the evolution of body cam video.
And it really has put it in extreme fine point on it
about how critical hyperscale computing is in particular for
the law enforcement mission.
>> Why do you say that?
Why, what do you see about body camera that is different in
kind?
Is it just a difference in scale that becomes a difference in
time or is there more to it than that?
>> There's much more to it than that.
So, it's a couple things, just to simplify.
One, it's just the sheer volume.
The data volume of digitized content produced in digital
cameras is significant.
And then once you store the data.
Storing it is a big enough challenge.
Can I rack spindles quick enough?
Then what are you gonna do with it?
So the ability to actually process it in a meaningful way.
And in many cases, 90%, I don't know what the percentage is but
at 90% body cam video or
more is probably relatively insignificant.
The day in the life of an officer
doesn't have events everyday, all day.
But 5%, 2%, even if 1% of that video captured,
generally speaking officers are interacting with people when
they're at their most vulnerable.
And the implications of that information and
data are quite significant.
So, the other implication is, what do I do with it?
How do I process it?
How do I share it?
How do I redact it?
How do I process it in some meaningful way?
How do I prove chain of custody?
If it shows up in a court of law.
So those kinds of things have really put a really strong
need for us to attack some really difficult problems that
people really haven't thought about before.
And, certainly providers like Microsoft,
it's a privilege to be, what we feel like is contributing
to the law enforcement mission.
And doing that in a responsible way.
>> So let me bring you into the conversation, Jenner.
And since we've gone down to the body camera question,
you don't have to move there, it's okay.
>> Want you to see eye ball to eye ball.
>> Yeah, okay, just in case the taser comes out.
I don't want to get in the way.
>> Yeah. [LAUGH] >> Since we've gone to the body
camera question, you know, tell us a little bit about taser's
development of the system.
>> Mm-hm. >> And
why it is that you concluded that cloud is the right answer?
>> Way to go? >> As opposed to stand alone
systems.
And what led you to find a cloud infrastructure partner?
>> Yeah, absolutely.
Thank you for having me here, Paul.
I'm glad to be here with Stuart.
Thank you for all your time for those who are joining us.
Maybe as I answer that question, I'll give a little bit of
a history lesson for Taser, and how Taser entered the body
camera market and the cloud services in general,
cuz normal folks out there that aren't in law enforcement
dealing with body cameras right now today, might not think of
Taser and cloud computing in the same sentence, right?
Taser was founded in 1993.
We're all familiar with the Taser devices that law
enforcement uses.
Raise of hands, who's been tased before?
I know we've got some law enforcement folks who have done
it as part of training.
Stuart, have you ever been tased?
>> Not yet. >> No
wild Friday nights that have resulted in a tasing?
>> Not yet.
>> No? >> Not yet.
>> Okay, all right.
[LAUGH] Around about 2006 or
2007, Taser started putting cameras on the bottom of
the taser devices, right?
Turned out to be very useful to have video of the incidents
of when a taser needed to be used,
when use of force had to be used by an officer.
And over time,
that quickly turned into the concept of body cameras.
Well, wouldn't it be nice to get a broader perspective of what
has happened in an incident between law enforcement and
a citizen.
And body cameras was the solution that Taser
started making probably in first 2008 or nine,
is when the first body camera prototypes came out.
Technology terms, that's like a eon ago.
[LAUGH] Right?
Quickly, the use of these early body cams demonstrated
the necessity for large backend storage management for
this digital evidence.
Not many law enforcement agencies were going to be
able to build their
own on premise infrastructure to manage that amount of video.
And we knew it was only going to get bigger.
So, pretty early on TASERbuiltevidence.com,
cloud-based digital evidence management system.
Probably launched first in 2010,
I believe, to go right along with the body cameras.
It was always meant as a cloud-based digital evidence
management system for our body cam videos,
any other body cam videos, any other source of digital evidence
that needs to be managed.
Interesting part of the story is early on, 2008ish,
Taser's first foray into building this backend management
system was an on premise system.
It was cloud based from the perspective of our customers but
Taser was going to build and run that infrastructure
in data centers that we, spaced that we leased.
As you can imagine that turned out to be difficult to do.
So it wasn't very many years into it where Taser made
a strategic decision.
We are really good at dealing with law enforcement.
We're really good, at this point, writing software and
building hardware for law enforcement purposes.
What we are not good at is running data centers.
Running racks and networking and cooling and
all those kinds of things.
So we pivoted to use infrastructure as a service
providers right about 2010.
And over the years we've used a number of different
infrastructures of service providers around the world.
Evidence.com is global, not just in the US, right?
There's deployments in Australia, Brazil.
Canada's coming, the UK, etc.
And we've benefited greatly by leveraging
infrastructure service providers such as Microsoft Azure,
which is our most recent partnership that we're
very excited about on running evidence.com on the Microsoft
platform around the world.
That's led to massive increases in our efficiency, right?
There was a large team of people running the evidence.com
infrastructure back in the day.
Now there's probably four or
five operational folks that can run all of evidence.com from
a daily operational perspective because we're leveraging
Microsoft Azure and other infrastructure providers.
Which is amazing when you think about it.
We can service law enforcement globally, 5000 agencies, over
two pedabytes of data store, a new video coming every one and
a half seconds with a handful of folks.
All because of the economies of scale that come
with hyperscale cloud computing.
>> Let me actually ask about that cuz you just threw out
a couple of numbers, and I want to, so it's a new video every
one and a half seconds that's >> That's you're global?
>> Globally. >> Global uptake to
evidence.com.
>> Correct. >> Right.
>> You said 2.5 petabytes?
>> Over 2 petabytes to date.
>> To date.
And this is very early in the body camera market growth.
>> Right. >> Body cameras are in the news
all the time, it's still very early in the deployment.
It is still relatively small numbers of police departments
have deployed these.
A lot have large agencies have.
But there is a lot of mid-size and small agencies that over
the coming decade are going to do that.
>> Do you have any estimate of what you think the average
officers upload per day, per month, per week is?
>> Yes, the average is about 1 gigabyte per officer per day.
>> 1 gig per officer per day.
>> And of course, that's not recording their whole day.
That's a couple, two or three contacts a day that
are recorded, results in about a gigabyte per officer per day.
>> Can you foresee any institution of any size,
larger than a six-person sheriff's office?
Person in Macon, Georgia or something like that.
Anything larger than that capable of actually doing this
on site, or is cloud inevitable?
>> From our perspective, cloud is inevitable, and
I have two really good examples.
Two of the largest police forces in the world are using our
solution with the cloud backing, the Los Angeles Police
Department, LAPD, as well as the London Metropolitan Police.
So the economies of scale actually get more
compelling the bigger you are, due to
the large number of storage arrays that need to be built and
managed, and the efficiencies and the solution itself.
Right, you're trying to manage that number of cameras and
that number of videos coming in every day.
It becomes almost impossible without a purposeful
automated system.
>> Let me bring it back to you, Stu.
You talked earlier about wanting to build systems
at Microsoft that were compliant with Segus.
Right now Segus is mostly about federal data coming
down from the FBI.
Do you manage when, is the Azure deployment that you're using
Features compliant, and why do you do it that way?
When you can maybe,
get away with abiding by Arizona law or Utah law?
>> Well it's actually funny you should
frame it in the context of law enforcement with get away with.
>> Well [LAUGH] it's totally fair.
>> Less onerous deployment required.
>> It's totally fair.
It's actually the appropriate way to frame it.
Because we're not interested in getting away with anything.
We're, in fact,
interested in helping move the industry in setting a standard
that we think is appropriate for a particular level of data.
Now, the interpretation of the Segus policy,
we could spend a whole lot of time slicing and dicing and
say this type of data requires this kind of requirement, and
this kind of data's this kind of requirement.
First of all, just for us from a scale standpoint,
this whole segmentation idea is incredibly challenging.
We decided that doesn't make sense to put your secure
data over here, put your non secure data over there
because if you could build the system and
make it cost competitive and compelling.
And your data isn't secure or isn't at high security
requirements, and you were still doing security.
Wouldn't that be better?
Because in the case of video or police data in general, a lot of
it might not be Segus data, but it's particularly sensitive.
And you know, when I go talk to the law enforcement folks,
I say things like, we have a conversation.
Segus data which is primarily background data just
oversimplified background data coming out of the FBI system.
Do you have any other important data?
I can tell you there's not an agency in the country that
wouldn't tell me that their payroll data is the most
important true story.
It's a very true story.
So we said, well, hey if this policy is good for
Segus data, it's also good for payroll data.
So when we look at our system we don't think about segregating
data we just provide that level of committment to customers.
Now there's some other standards that we also address,
HIPAA, for people who are familiar with healthcare,
IRS-1075 is really kind of a financial data kind of a thing.
We apply those standards to all the data.
So even though body cam data might not be IRS 1075 data,
you can rest assured that those same standards
in IRS 1075 are being applied to the video data.
So I don't know if that answered your question, but
it was really our intent,
because no one had really addressed that before, to
figure out how can we stand up and responsibly make commitments
to this kind of data, and some of the things specifically.
Around Segus are background checks for our employees.
I happen to be one of them.
I'm not in the data center but I decided hey this is going to be
a good, this is a good idea for Microsoft and our employees,
I'm going to volunteer.
I spend a lot of time with law enforcement
across the country so I raised my hand and I'm fingerprinted,
background check in over 17 states currently.
It's kind of the same background check, somewhat redundant.
But that's one, the background checks.
And that's hard.
It's really hard for
service providers whether it's a hosting company or
a service provider.
The second thing is audits.
How do you do audits?
Which we could maybe talk a little bit more about.
And then the third thing is just the contractual commitment.
Actually how do you contractually commit to those
policies?
>> I'll jump in here real quick.
And certainly the siege of security policy is of critical
importance to our customer base.
As we deal with the body cam videos in a cloud scenario.
And so partner with Microsoft, one of the Major benefits to us
is their commitment to the law enforcement market and
the Segus security policy, chasing down all those details
that [LAUGH] Stuart was just mentioning there, right?
Taser's not a huge company.
There's maybe two or
three of us in Taser that deal with the Segus security policy
to ensure that we're meeting it.
To spend time going state by state to understand each state's
unique requirements and help get the agreements done and
the background checks done in each state.
And so it was critical for us to have a partner that can
essentially use bigger resources and
go take the lead on some of that stuff to help us
navigate those waters because we're also doing it globally.
We're not just- >> So for
you Seguis is just essentially a turn-key operation.
>> Well at least it's a strong foundation.
There's certainly a lot of Segus elements that we do within our
program for evidence.com specifically but
it is a very solid foundation to work from.
>> Actually you mentioned the globalization a couple times,
and I am sure most the people here
are probably domestic, because.
>> Yep.
>> Obviously we are here in New Orleans and
it's unlikely that too many people came from Australia.
But your perception on whether or not the standards are harder,
easier to meet overseas, different.
And how the Azure deployment kinda matches up
to the others as well.
>> So just different, meaning I think from a fundamental level,
the security requirements from frameworks around the world
are very, very similar.
They're requiring the same kinds of things.
Background checks, encryption arrests,
encryption transit, vulnerability management,
penetration test, a lot of the exact same thing.
It's just using different language.
Different frameworks and structures.
One of the benefits we are getting from Microsoft
partnership is folks like Stuart are helping us chase down
Segus in the US but
we have other Microsoft partners n Australia and Canada helping
us chase this perceivable requirements in those areas.
Although they may b very similar in operational and
he the we execute, they are different in language,
different in implementation,
different oversight bodies that we need get approval from.
And so have a global company help us in
all of those markets around the world has been fantastic.
>> Back to you, Stuart or maybe, maybe you Jenner, I don't know.
One of the things that you talked about was the value
of not having two buckets of data,
kind of low secure and high secure.
Do you have any estimates, have you done any work on figuring
out what the kind of savings in what I would call curatorial
costs, the idea that somebody has to manage which is which,
and has to move low secure to high secure and
back and forth, and make those judgements.
Is that an expense that is being kind of taken out of play
by the high secure standards, and
do you have any sense of how much that is?
>> That's actually a great question.
And now that [CROSSTALK] you brought it up, it would be,
it would be really compelling if we could sort that up.
I can tell you about one agency that I was recently working
with, they were looking at deploying a five thousand,
excuse me, $5 million implementation on premise.
They can do it in the cloud for 50,000.
And that's just the act, I wouldn't say it's impossible,
yet almost impossible to articulate and quantify.
The $5 million investment also included a whole bunch of
procurement in the front end, then they would have to rack and
stack it, they'd have to get it up and running, and
all that whole process.
All that process.
I It would be quite significant, so not only is the initial
financial outlay cost pretty significant, but to your point
the cost behind the scenes are also fairly significant.
The other thing I wanted to say, I didn't get the chance earlier,
related to the kinds of data and segregating data,
we just announced a couple of weeks ago, FedRamp pie for
our Azure environment.
And not to get too much into the details cuz the compliance guys
like Jenner and I get into the acronyms and We'll start talking
missed 853 or something- >> [CROSSTALK] Stuart's the only
one that likes to talk about that stuff.
>> [LAUGH] >> We just announced
FedRamp High which is really the highest level of the federal
government's look at us.
And then we also announced Odessa 4 and Odessa 5 which is
kinda a segment in environment, really more about DOD.
But if we drew a timeline and we look at the kind of commitments
we've made over time, we're continuing to increase and
add those things.
There's not a arguably a government at the local level or
a law enforcement agency that has actually FedRAMP High
requirements, because these are really the things that reach up
into the national security realm.
>> Mm-hm. >> But that's the same
environment that we have our customers run and operate in.
So that's how we can kinda stand up with this level of confidence
about, hey, look, we not only you think you can trust us but
we're gonna prove it.
>> And maybe a little perspective here from
a infrastructure provider standpoint,
like Microsoft Azure, it makes all the sense in the world to do
set up the data center at the highest standard possible so
that you don't have to worry about which data can go in,
which data can't and segmenting data from that level.
When you take one step down to the software as a service
provider, right, we benefit from that cuz it doesn't matter what
data we're working with.
We know on the infrastructure level, its at the highest
level possible, but then it still makes a lot of sense
on our end to still have some kind of data classification for
how different levels of data's protected,
even down to the customer level, right?
Cuz there are costs involved if you decide to protect everything
at the highest level, well, now you're spending a bunch of
effort on something that might not be necessary.
But cloud computing helps enable a more rational structure.
[CROSSTALK] More agile way for
how you might make those decisions at what levels,
on how you segment and control different levels of data.
Whereas doing different types of security for different data
within an on premise environment can get very complex,
cuz now you're duplicating infrastructure.
In a virtualized cloud environment,
it's a lot easier and
costs a lot less to do a more refined segmentation of data and
have different controls applied separately.
But certainly from a fundamental infrastructure, yeah,
protect it at the highest level possible,
then everybody downstream benefits.
>> So now let's pivot a bit from our discussion.
We've been talking about the cloud and
the security of the cloud.
I wanna now turn to kind of user service and functionality, and
some of the things that are coming down the pike in the way,
right now, we're just at the beginning of
body cam deployment and we're getting reams and reams of data,
terabytes, petabytes of data every day, week, month, year.
A lot of that is going to require analytics
at the back end.
Things that are going to make this data useable so
that you can find what you need and either get it to court or
get it to your intelligence community or
within the law enforcement, wherever.
So, couple of the things that occur to me as kind of important
functionalities and want either talk about how they're working
in body camera and how they're coming through in the Azure.
Let's start with chain of custody, right.
You got a body camera, got a video, it's got evidence or
potential evidence, criminality,
eventually it's going to have to be tagged, maintained and
segregated, and then if there's a court case, used by the court,
used by the prosecutor,
eventually discovered to the defense attorney, etc.
Tell me about that functionality,
where you see it going in the future, what you see coming out
of the body camera, and how it's gonna work in Azure.
>> Sure, sure. I'll start with how the old
school methodology of chain of custody for video evidence.
You can imagine an older dashcam system or maybe the video was
burned right to DVD in the car, they take the DVD out,
walk it into the police station, put it in the evidence room.
You have stacks and stacks of DVDs.
Time to take it over to the prosecutor, go find the right
one, put it in an envelope, drive it over in a car.
So how do you keep,
maintain chain of custody in that scenario?
Well, you write down who took it when. Okay?
And there's also all
the opportunities in the world to mess with that data.
It's not terribly well protected in that case scenario-
>> Or lose it.
>> Or lose it altogether.
>> Or misfile it.
>> Exactly.
In a cloud scenario, all of a sudden you have immediate
benefits because the video now goes from camera to cloud.
It's in the cloud system.
And every time somebody touches it, looks at it, downloads it,
shares it, all of those actions are logged and
can always be proved cryptographically with
hashing functions that the file that ended up in court is
exactly the same file that was on the camera.
Even if that was years and years before.
That's a quantum leap forward in chain of custody capabilities
from the way we've been doing it in law enforcement for
a lot of years.
And that can only happen in a cloud scenario.
You couldn't replicate that kind of chain of custody having
independent on premise systems of digital evidence management.
A centralized cloud-based evidence management system's
really the only way to have a really tight chain of custody
from camera to courtroom all the way through.
>> The other thing that I would just add that is going to change
significantly in the process that Jenner just described
which is too familiar, and
it's not that people have made bad decisions.
People are just- >> That's all you could do.
>> It was the best you could do.
The idea of the DVD coming in and out.
That's really, that was state of the art not that long ago.
But one of the issues, what he just had mentioned,
when you're signing something in and out, historically we've had
this trust of a wet signature and what does that really mean?
Who really signed the piece in?
And you look at this signature and is that?
In the digital space, we have digital signatures or
digital authentication.
We're kind of at the cusp of that as well.
Meaning, we can talk about multifactor authentication,
different identities, but the idea of sharing data within
the cloud, you also have another level of what I would say
confidence that the data that you shared or
who touched it or who had access to it or who modified it
is actually auditable at even a higher degree of standard
than trying to find this piece of paper with a signature and
validate that it's a legitimate signature, if that makes sense.
I know a lot of the work that they've been doing and
you have a repository and you've shared data with a prosecutor,
a defender, whoever, you can actually authenticate and
tell who accessed what data when.
And then you can stand in a court of law and
actually with a really higher degree of certainty actually
validate that in fact someone had accessed to data or not.
>> So, you envision a functionality that essentially
has proof positive identification of all
digital manipulation of the data from creation to-
>> Yeah, very much so, and
not only that but the ability to share across jurisdictional
boundaries in ways that probably weren't possible before cuz in
Jenner's example, the DVD, do you make a copy of it?
Two or three people need a copy.
Then you gotta burn copies and
then you gotta distribute those copies.
You don't have to do that in a cloud space.
You can distribute digitally and actually use this authenticated
directory structure so you know exactly who between different
jurisdictions has been- >> [CROSSTALK] Not to mention
how the sharing happens and
controlling what the receiver's able to do.
They can only view.
They can download.
They could re-share or they can't, for how long.
The kinds of controls you can now put over this
management of this evidence that couldn't possibly do in
the physical world.
>> I actually I was struck by the fact that we won't lose it.
I don't know if you've read it, but
they've just found Wilbur and
Orville's patent application after it had been lost for
40 years cuz it was misfiled in the archives.
So we won't have that problem anymore.
>> [LAUGH] >> Mm-hm.
>> But okay, so now we've got digital data and
we can decisively know who's touched it.
One of the reasons people are gonna touch it,
though, is actually to edit it, to redact it,
either to edit out of the officer's casual conversations
about what a jerk the chief is, right?
[LAUGH] Or- >> Officers never talk that way.
>> Never talk that way, I understand.
Or more realistically, to edit out or
blur the facial features of innocent bystanders,
witnesses whose identity hasn't been disclosed.
Tell me about that.
How are we gonna manage that?
Is that going to be subject to the same sorts of authentication
requirements and
how do you see the technology developing to make that easier?
Cuz that strikes me is going to be massive
part of the problem going forward.
>> Yeah, so the issue of derivate works,
snippets of video, redacted video or
whatever, that's fairly easily solved technologically,
you can always create, downstream works that can be
still cryptographically tied back to the original.
And you can prove in court that this ten-minute segment of
video, we can authenticate that it came from this four-hour
piece of video, if you ever wanna validate that back.
That's already exists in evidence.com today.
That's not terribly difficult.
The much more difficult scenario is redaction in general, right,
where, how does law enforcement now that they have these mounds
and mounds of video quickly and easily redact appropriately and
100% accurately?
You make a mistake on redaction and that's got to be accurate
and be able to do that efficiently enough for court use
or public disclosure or release to the public or whatever it is.
And so, we're working on tools to do that.
We have processes right now at evidence.com that are pretty
good, not completely fully automated.
And I think over the coming years,
you're gonna see people like Microsoft probably accelerate
the ability to do automated redactions, following faces and
objects and things like that,
at a much higher clip than what we can do right now today.
That's certainly in the next wave over the next year
or two I think.
>> Let me maybe step back a little bit and
just give everybody a frame,
cuz everybody's not in law enforcement and
what we kinda get is the headlines. Right?
The sensational headlines on
why body cameras, not body cameras,
all of that, the politics of body cameras.
So let me get out of the politics of body cameras.
Let's talk about the day in the life of an officer and
the kind of information that may be captured.
Imagine an officer gets a call for
a child abuse case to go investigate.
And she arrives at a hospital cuz that's where the child is.
She gets out of her car, walks into the hospital,
what happened?
Is the camera on, is it off?
When she walks in, what about HIPAA?
What about all the other people in the hospital?
She walks into the room, and there's a child.
She's interviewing a child.
Is the camera on, is it off?
There's a mother there and
the mother has a court case against her,
and there's a garnishment of wages and finances come into it.
Is the camera on, is it off?
Did you just capture that financial information that might
be IRS 1075.
The uncle shows up, the uncle wants to help the kid and
the point is the officer, at that point in time, is just
trying to make a decision about the safety of this child.
Should the officer be worried
what kind of data is being captured?
How to route it, where to route it?
So that kind of goes back to my opinion of,
or our opinion of, protect the data.
So if you decide to stream live it on YouTube later,
which one agency determined to do,
not that particular data, but they were going to make all
their body camera data live on YouTube.
But that is okay, it's kind of one extreme.
And we've had other agencies that said no other
data is going outside ever.
The answer is probably in the middle, the one that went live
on YouTube, realized they had do a lot of redaction work.
And what they ended up publishing on YouTube was
just a blurry image with blurry garbled vocabulary and really
wasn't that much value so- >> The only way to do it
efficiently enough.
Just redact everything.
>> Just redact everything.
>> So the idea of what they were publishing didn't really meet
the value of transparency.
So what we're gonna see in my opinion, we're gonna
see happening is continued evolution of this technology and
it comes back to what I said initially.
The capabilities of hyper scale computing are enabling
a new generation of tools of what these are,
these are public safety tools.
And there's a huge benefit, a huge benefit to body cameras for
example, for officer safety.
There's actually some statistics out there that show that it
turns out that not only will an officer's behaviour
change which is kind of the politics of it, but it turns out
that people on the other side, their behavior changes.
And there's a lot of evidence that can be captured
by body cameras.
This technology and the ability to apply what's really
an unlimited amount of CPUs, that's what hyperscale is.
You have an unlimited amount of processing power and you have
good software you can actually do some really interesting,
creative things, we're very much on the early cusp of this.
Jenner mentioned we are doing all kinds of things in our
research and development department, with Azure Media
Services and a lot of other research capabilities that allow
parsing and dissecting video in creative ways.
Identifying things, being able to automate, it's not perfect
today, there's still a lot of work that has to be done.
And the video that I mentioned of the officer in the hospital,
there's probably not a future where that's gonna be 100%
computerized, I don't think, in spite of the talk of artificial
intelligence and all that.
[CROSSTALK] >> Right.
>> There is still going to be a
requirement for people that can interpret social interactions,
or however you describe it in some meaningful way.
But we're also creating tools
for them to that.
>> Exactly, then it comes down to the workflow opportunities
and intelligently picking out when a human needs to go in
there and make those decisions.
One thing I want to point out when it comes to developing some
of these more advanced technologies like
facial recognition and automated redaction.
The cloud services make it incredibly flexible for us as
a company to decide are we gonna use Microsoft provided facial
recognition tools that could be part of the Azure deployment.
Or, if we think we can do it better, great,
we can still write it ourselves, do it better, and
just make use of the computing power.
>> That's right.
>> It's really nice to have that flexibility to
leverage the kind of advances are coming or do it ourselves,
and whatever makes sense.
>> So pull back the screen, we've got a little bit of
time left, and now peer into the future five years.
And you mentioned facial recognition,
I could also imagine object recognition.
>> Mm-hm.
>> Tagging so that video is automatically catalogued and
tagged so that it recognizes
where it was taken, tied up to a geo location tag or
recognizes the nature of the events there.
And so this is a traffic stop, this is a homicide
investigation, this is a, so imagine for me what you see
coming in terms of the software pieces that are gonna lie
in the future behind the prevalence of the data.
>> Yeah, so- >> Or for
that matter the video pieces here,
if you're thinking of deploying infrared or something like that.
>> So along the lines of data management,
what's happening already today,
the capabilities that already exist today is to easily with
a mobile device or with integrations in the CAT or RMS.
These are law enforcement data management systems,
to tag the videos that are coming in based on what type of
call it was, was it a traffic stop,
was it a man on the street interview, whatever.
That can already be done In a somewhat automated but
mostly manual fashion today, which then allows a little bit
of backend analytics and quickly sorting and searching and
finding what you need to find.
What will be coming next is automating those processes even
further, based on the content of the video,
I know this was a traffic stop.
That might be fundamentally one of the easier ones to determine.
>> Right. Those up to
the side of the door.
>> Yeah, but you can see where it's video analytics get better,
you can automatically determine what type of situation it was.
Which just makes the officer's job easier when they're trying
to tag in their information so that those upstream can find
what needs to be found and can used the cases appropriately.
>> So let me say it this way, Paul.
The future is already here.
>> [LAUGH] >> I can tell you one thing,
we are going to woefully overestimate,
I'm stealing Bill Gates' quote by the way, we're going to
woefully overestimate what will happen in two years.
And totally underestimate what will happen in ten.
And so, this would be fun because I know this is being
recorded, ten years from now, someone is gonna watch this and
say, those goofballs had no idea what was happening in ten
years and overestimated what happened in two.
>> But give me your best guess in ten years, shooting high, so
that Bill Gates is proud of you.
>> [LAUGH] I'd say a couple things.
One, first of all, I think it's important when we're talking
about body camera video in particular to get our head
around that there's a lot more than just body camera data.
The ability to aggregate multiple,
which you guys are doing a great job, of multiple data sources.
So whether it's a fixed camera at the bank,
respond to a bank robbery, you've got body cameras on-
>> [CROSSTALK]
>> You know what,
there's a whole bunch of video on bank video.
>> Citizen cellphone videos.
>> Citizen cellphone video.
Location awareness, GIS, so
the idea that the camera's location aware.
So being able to aggregate that into one
set of information around an event as opposed to an object.
So this data came from the body camera, rather than that,
this data came from the bank incident, all in one and
created a whole evidence system.
That's what these guys do as a profession.
The second thing is the idea of, you mentioned infrared,
but cameras are gonna get smarter and smarter and
faster and faster.
All it really is is a sensor, I mean if you
take the word camera out and think about the word sensor.
What other things can it capture?
There's a whole bunch of audio.
Could a camera be smart enough to know when an officer says
specific words that does specific things, like,
I'm in trouble, dispatch knows, and somebody is on its way.
A gunshot, hm, is it difficult for
a computer to understand a gun shot?
In some scenarios, yes, but in most scenarios, I would argue,
we can probably do that now.
>> Mm-hm.
>> What we'll be able to
do in a few years is actually do something interesting with it.
Third thing, will cameras that are around each other, know each
other, will they be self aware about other cameras in the area?
So a camera, one officer, for example,
might have his camera on, cuz he's in an incident or whatnot,
there's another camera nearby.
It automatically gets activated because it knows something's
happening and can you triangulate.
Can you do a three dimensional rendering?
Where we use crime scene photos today,
imagine if we had a three dimensional rendering.
And we can get way out there and
think about things like HoloLens.
>> You're kind of giving away our entire product roadmap.
>> Sorry. I'm just telling
you where the technology's going if you've seen HoloLens there's
this idea you have this virtual reality environment.
So instead of a two dimensional rendering of a crime scene,
you could have a three dimensional rendering,
which allows a whole bunch of other sophisticated things.
>> The Oculus.
>> Well, the HoloLens in our case.
>> The HoloLens, sorry, sorry.
>> People might be interested to know what is actually can
happen today.
>> So right now, in the TASER ecosystem, you
flip a safety on a taser, the cameras in the vicinity go on.
In the cars,
you turn on the lights, the cameras in the vicinity go on.
Coming soon would be things like drawing a gun from a holster,
cameras go on.
Not that many years down the road it's gonna be some of these
other sensor, including maybe vital signs of the officer.
>> Yeah I was gonna say- >> Heart rate goes up.
>> Yes, heart rate goes up or he's running.
>> He's running.
>> [INAUDIBLE] location he's running that means something is
happening as opposed to walking.
>> Yep, exactly.
>> That'd be fascinating.