start	end	text
1120	3020	Oh, I guess it is.
3820	5180	I guess it is.
5180	7060	There we go. Awesome.
7060	9500	Because as we were just sitting here joking,
9500	11260	we're like, uh, this is absolutely
11260	13060	the slot to see who
13060	14660	made it to bed early.
14660	17020	And who has already, ooh,
17020	18600	and now I'm everywhere.
18700	20500	This is awesome.
20500	23360	We called it the hangover slot.
23360	24660	So welcome.
24880	27080	Yeah. And of course we only
27080	29040	have one microphone, so the good
29040	31240	news is that works better.
31320	33300	So welcome, everyone.
33640	35100	And we're going to take a little
35100	37000	bit of a different approach,
37000	39200	I guess, than just presentation
39200	40640	slash panel.
40780	42780	Because one of the
42780	44900	issues with this is
44900	46940	it's an evolving space when we're
46940	48760	talking about the data and the privacy
48760	51020	and all the issues with,
51020	53240	I mean, if you ask anyone to define
53240	55480	as Britton's going to talk about,
55480	57400	like, some of the aspects
57400	59540	of the data privacy in XR.
59540	61700	I mean, we don't even have a common definition
61700	63660	for some of the technology,
63660	65620	so then how are you going to define what it's
65620	67960	doing? So I'm Liz,
67960	69680	and this is Britton, and
69680	71660	we're going to talk, first Britton's
71660	73480	going to lay a great foundation
73480	75300	and give us a framework
75300	77280	from which to have a discussion.
77300	79620	And then the idea is we'll talk
79620	81460	amongst ourselves, and then once y'all have
81460	82660	had an opportunity
84360	86480	to wake up,
86480	89300	get the brain cells firing,
89300	90600	then we thought we would open
90600	92520	it up to a conversation and
92520	94780	questions from the group.
94780	96420	So, bear with us,
96420	98600	and I'm going to let a much smarter person
98600	100520	than I introduce the
100520	101620	topic and
102420	104480	explain a little bit of where some of the
104480	105720	challenges lie.
108740	111320	Question one, can you hear me?
111600	112560	Awesome.
112620	114600	Question two, who's heard
114600	116080	of Lego Universe?
117540	118620	Alright,
118620	120420	I have a salty story
120420	122300	for your Sunday morning about
122300	124390	Lego Universe.
125120	126320	Lego Universe
126320	128100	was a highly
128100	130220	anticipated massive multiplayer
130220	132600	online game, like Roblox,
132600	134360	Minecraft, that
134360	136250	ended up being shuttered.
136960	138320	The experience looked
138320	140280	great, and the company
140280	142320	was really looking to allow
142320	144480	users to build their own type
144480	147100	of content and their own environments.
147400	148540	But given how utterly
148540	150840	trusted Lego was as a
150840	152700	brand, they committed to
153500	155580	whitelisting content before
155580	157800	it went live. So they were going to preview
157800	159340	everything to make sure it was safe
159340	162460	for you and your kids.
162460	163620	The only problem was that
163620	167240	there's a term in content
167240	167700	moderation that
167700	169460	Lego did not think about, and
169460	172240	it's called time to penis.
170480	172480	Time to penis
172240	173860	literally means
173860	176100	the duration
176100	178040	between a person entering a virtual
178040	180260	world and adult content
180260	183400	appearing in that virtual space.
183500	184280	And with all puns
184280	186120	intended for Lego Universe,
186120	188560	the time to penis was not very long.
190520	192280	Lego needed, amongst
192280	193660	other things, to
194240	196920	figure out an automated,
196920	198500	scalable, economical
198500	200520	way to prevent teenagers
200520	202840	and kids from drawing penises
202840	204580	in the virtual world,
204580	206480	and they couldn't do it.
206740	208560	It became a wide-scale
208560	210760	problem, and it actually ended up killing
210760	212200	the project.
213220	214600	This is not an
214600	216780	unusual story. Many
216780	218520	companies find that their expectations
218520	220700	about governing and moderating
220700	222460	user behavior and
222460	224560	user data flows and spatial
224560	227440	computing are turned on their head.
228440	228740	My
228740	230560	name is Britton Heller, and I
230560	232600	like to describe myself as the lawyer
232600	234120	of the black mirror.
234760	236680	I am a visiting scholar at the
236680	238600	Virtual Human Interaction Lab at Stanford
238600	240500	University, and I teach
240500	243060	international law and AI policy.
243060	244820	And I've been focusing on immersive
244820	247060	environments for about seven years.
248260	250700	There are three things that I would like you to
250700	252060	take away today.
252400	254680	One, immersive media
254680	256840	has a different impact on our
256840	258970	bodies and our minds.
259680	260880	And this is what
260880	262980	differentiates it from flat-screen
262980	264700	social media.
264920	266680	Existing privacy law
266680	268860	does not accommodate for
268860	270080	this.
270600	272600	Two, the different technical
272600	275160	stacks between flat-screen
275160	277000	social media and
277000	278940	immersive content
278940	280840	mean that we have different
280840	283020	layers of enforcement when you're
283020	285120	looking at behavioral tracking
285120	287260	and modification in XR.
288020	289180	And three, AI
289180	291260	will not save us. And that one is
291260	292920	worth repeating twice. AI
292920	295300	will not save us.
295640	297200	Overall, the systems that we
297200	298800	create in 3D environments
298800	301020	are fundamentally different
301020	302780	from 2D environments
302780	305020	because of the technical stack, the
305020	307000	potential impact on users, and the
307000	309280	uninformed state of the law.
309720	311020	This leads us to three
311020	313080	questions. What kinds of data flows
313080	315520	do you get from XR environments?
315520	316920	What does this mean for business
316920	319000	models and the law? And why does
319000	320500	this matter to users?
323180	324780	Imagine that you and I
324780	327160	are playing a race
327160	329560	car-oriented game in XR.
330200	331180	And I see this
331180	333180	red McLaren, and I
333180	335160	really, really like this car.
335160	337120	I was going to say a Tesla, but the one that they're
337120	339260	hacking over there, I don't know if it's actually going to be able to drive
339260	341280	after this week. So, I
341280	343020	like the Tesla. I like
343020	344760	McLaren. You can like the Tesla.
346340	347220	When I see
347220	349360	this car, my body starts
349360	351220	to physiologically react
351220	353060	to my pleasure. So, my
353060	355160	heart rate goes up a little bit.
355160	357260	My skin gets a little more moist.
357260	359060	My pupils dilate.
359180	360860	I really like this car.
361860	363180	Later on, as we're walking around
363180	366060	the conference, and we go into a different experience,
366060	367260	I see
367640	369160	a similar red car drive
369160	371300	by, driven by someone who looks
371300	373180	more than a little bit like me.
373980	375200	I go to a virtual sports
375200	377580	bar, and someone who looks exactly
377580	379060	like my crush
379060	381160	says, isn't it a great
381160	383340	time to buy a new car right now?
383800	385200	I go to my email, and
385200	387320	I start getting targeted ads from
387320	389260	Geico about why I should
389260	391180	re-up my car insurance.
391980	393240	This sounds like
393240	395520	science fiction, but the framework
395520	397180	for it actually already
397180	399180	exists. Companies have been
399180	401640	putting advertisements in XR environments
401640	402460	since
403940	405440	2021, I believe
405440	407200	it was announced. And I've
407200	409240	written papers about how advertisements
409240	411200	in immersive contexts are
411200	413180	actually experiential.
413400	415240	They don't look like billboards.
415240	417320	You actually pay for the privilege of having
417320	419400	the car advertised to you.
420380	421440	I developed
421640	422740	a term called
422740	424480	biometric psychography
424480	426140	to describe
426140	428920	the potential for this harm.
429300	430240	Where your
430240	432540	likes, dislikes, and consumer
432540	434580	preferences are married
434580	436100	with your physiological
436100	439680	reactions, often involuntary.
440280	440860	This is
440860	442240	not against the law.
442620	444560	This is not covered by normal privacy
444560	446520	law, because privacy law covers your
446520	448880	personal identifying information.
449420	450600	And when you and I entered into
450600	452640	the game, I had a verifiable billing
452640	454680	address. I had a user account
454680	456620	with my authentic identity.
456620	458560	The companies know who we are. That's not
458560	460480	what's at issue. It's the
460480	462880	inferences about our behavior
462880	464220	and our physiology
464220	466520	and our preferences that are really
466520	468560	at risk. So,
468560	470540	it's your mental privacy, not your
470540	472800	identity. So, privacy law
472800	474600	is not the right regime to deal
474600	475660	with this.
477680	478220	It
478220	480500	actually makes me a little bit upset, because
480500	482480	I'm a lawyer, and I'm like, I don't
482480	484600	like it when new technologies
484600	486100	are a misfit with
486100	487640	existing law.
493140	494720	Because there are different
494720	496600	psychological and neurological
496600	498520	impacts than
498520	500280	socializing with people in video game
501040	502240	worlds,
502240	504540	I'd like to review the three characteristics
504540	506940	that make something immersive.
507060	508540	And I think that when you understand
508540	509720	this, you
510460	512420	really get that this is not
512420	514380	flat screen media anymore.
514860	516720	First is presence.
517260	518520	Presence has been studied
518520	520360	in communications departments for
520360	522420	decades, and formally it's
522420	524710	called the illusion of non-mediation.
525740	526380	When you go
526380	528680	into an XR environment, it feels
528680	529760	real.
531200	532540	The first thing
532540	534500	that we do at the Virtual Human
534500	536440	Interaction Lab when we have somebody visit
537100	538460	is I have them walk
538460	539540	the plank.
540200	541920	I put them in an experience
541920	543640	where they are over a chasm and they are
543640	545640	walking along the plank because nothing
545640	547520	brings the feeling of presence
547520	548500	to you
549280	551760	more acutely than feeling that
551760	553760	vertigo and feeling your
553760	555560	heart rate rise and seeing how your
555560	557540	physiology reacts
557540	559540	to the screens and the
559540	561260	spatial audio and the impact
561260	563600	to show you that this is not a Twitter
563600	565700	feed, this is not Facebook, and therefore
565700	567660	maybe we shouldn't be regulating it in the same
567660	568460	way.
572230	574190	Jeremy Bailenson, who is the founding
574190	576290	director of the lab, says we should think of
576290	578330	VR not as a media experience
578330	580330	but closer to an actual
580330	582370	experience. And the
582370	584170	visceral pit in your stomach
584170	586150	you get when you have
586150	588570	fear of heights,
588570	590270	that's what
590270	591350	that is.
592650	594190	And we do that before talking
594190	596250	about empathy and education
597030	597630	and
597630	600030	all of the
600030	601830	other benefits of VR so that people
601830	603570	understand its power.
604010	606030	Second is immersion, and that's when users
606030	607990	really feel like they're in a virtual
607990	609330	environment.
609870	611970	A virtual violin performance has to
611970	613910	have all of the trappings of watching
614090	616290	a performance in an actual theater.
616470	618290	Different views from different seats,
618290	620390	different movement flows through aisles,
620390	622530	spatially oriented sound.
622670	623370	So it's
624630	625840	the way that
625840	627920	XR experiences use your sensory
627920	629960	input to stimulate the
629960	632190	user through light, sound,
632740	633780	sense, tactile
633780	635520	input.
636000	637820	Finally there's embodiment, and this is
637820	639860	my favorite one because it reminds me
639860	641360	of Looney Tunes.
642240	643840	Embodiment is the feeling
643840	646000	like your avatar or your virtual body
646000	648150	is your actual body.
648760	649880	And a researcher
649880	651040	named Mel Slater
651840	653740	took a famous social
653740	655440	psychology experiment called the rubber
655440	657720	hand experiment. And he had
657720	659240	people look down in VR
659240	661700	and see a rubber hand placed in front
661700	663700	of them, and then proceeded to go
663700	666200	Looney Tunes on it, to hit it with a
666200	668500	mallet, to stab it, to really
668500	670240	abuse the virtual arm.
670240	672020	And when you watch the recordings of these,
672020	674920	you can find them on YouTube, people
674920	677020	scream, people jump,
677020	678920	people react like it's their
678920	679440	actual arm.
681500	682440	This is the
682440	684400	power of embodiment, and it can
684400	686740	be beneficial in medical applications
686740	688460	if somebody has phantom limb pain
688460	690500	where you want to show them an arm that is
690500	692460	not theirs, and have their brain actually
692460	694040	associate what happens to that
694040	696020	with their physiognomy.
696120	698520	But it also can place users at risk
698520	700600	considering the violence
700600	702400	and other user harms that we're
702400	704820	starting to see emerge in XR.
706440	709120	So if this feels so real to our bodies,
709120	710920	what does it do to our minds?
712540	713600	Tom Furness,
713600	715020	who is one of the inventors of VR
715020	717100	headsets, says that when something happens
717100	719020	to you in XR, it's like it's
719020	720840	imprinted on your brain in permanent
720840	721780	ink.
722880	724960	The experiences that you have in a virtual
724960	727260	environment are processed through your hippocampus
727260	728740	in the same way that you create
728740	731120	memories, like you
731120	733240	and I sitting here having a conversation
733240	734100	today.
735840	736920	It's more like
736920	738880	entering somebody's living room
738880	740640	and screaming in their face
740640	743060	than it is reading critical
743060	744740	words on Facebook
744740	745960	or Twitter.
747440	748860	Because of how
748860	750980	XR impacts our bodies
750980	752960	and our minds, it's a mistake
752960	754860	to just cut and paste terms
754860	757100	of service from flat screen social
757100	758980	media and presume that
758980	761000	they're going to do the same thing. And that sounds
761000	762920	really obvious, but that's what a lot
762920	765060	of companies are doing these days.
766360	767200	Second,
767200	769240	I promised we'd talk about different stack layers
769240	771100	and moderation vectors in spatial
771100	772160	computing.
773360	775140	What that means in normal people
775140	777140	speak is that when you do content
777140	779300	moderation on Facebook or Twitter,
779300	780940	you look at two things, conduct
780940	782280	and content.
783080	785040	When you move from a 2D
785040	787140	to a 3D environment, you have
787140	789200	to look at content, conduct
789200	791100	and environment. So the
791100	793240	architecture of your world,
793240	794980	like we talked about with the
794980	797500	penises in Lego Universe,
797500	799700	becomes a moderation factor.
799700	801520	One that you don't have
801520	803440	when you have a text-based
803440	805380	social media environment.
807700	808900	That means that there
808900	810800	are actually different ways to
810800	813000	mess with people in
813000	814940	XR worlds. I like
814940	816560	to think about spam.
816680	818800	And how does spam translate into
818800	820960	XR? Pretend
820960	822880	you and I are tired of my
822880	824420	sports car, so we go to a sports
824420	826820	bar. And in the
826820	828460	sports bar, I can buy Liz
828900	830800	a beer. I don't
830800	832900	want Liz coming to my bar
832900	835200	though, or playing on my pool table,
835200	836900	so I buy her so many beers
836900	838920	that it fills up the virtual pool table
839420	840820	and she can't play the
840820	841540	game.
843260	844980	Impeding a user's gameplay
844980	846420	in an XR environment
846420	848600	is the functional equivalent to
848600	850660	spam in a flat screen computing
850660	852520	environment, but it would never be defined
852520	854760	as such if you didn't think about it
854760	856640	in three dimensions.
862160	863360	I'm going to skip a
863360	865160	little bit ahead because we have a lot
865160	869960	to talk about.
869960	871300	Three, platforms
871300	873200	are having a lot of challenges
873880	875340	trying to create trust
875340	877800	and safety regimes and privacy
877800	879640	regimes because of all of
879640	882480	these complicating factors.
882480	882900	This is a
882900	884660	new form of media with
884660	887040	new impacts on your body and mind, you need
887040	888400	new rules.
889340	890900	A lot of people ask,
890900	892920	why didn't Lego just create a filter
892920	894680	to screen out
894680	896180	all the penises?
897160	898780	And the answer was
898780	900640	that most social media content
900640	902860	is moderated by automated
902860	904840	filters in an ex-ante
904840	906920	fashion, so before it ever hits
907400	908580	a flat feed.
908720	910860	Facebook takes 97% of hate
910860	913120	speech off before you ever see it.
913420	914840	And the way that they do this is
914840	916700	by using classifiers.
917180	918000	So it's
918720	920920	basically a huge AI profanity
920920	921940	filter.
922840	924740	This is not possible
924740	927040	in the XR context yet.
928600	930460	We don't have
930460	933040	scalable, affordable,
935080	937180	reliable ways to create
938320	939000	classifiers
939000	940900	for 3D environments.
940960	942700	We don't have the data sets yet
942700	945140	to use. Computer vision is
945140	947060	the way that I think
947060	949080	we're going to be able to do this, but think about
949080	950900	all the different ways you can
950900	952200	describe a cat.
953080	954320	We don't have
954980	956920	publicly available data sets yet to try
956920	958820	to audit what a cat would
958820	961160	actually be in its many forms
961160	963500	and colors and shapes and sizes.
964700	966720	Currently when companies do content
966720	968900	moderation, one, they
968900	970640	just don't do it for XR.
970940	972880	Two, they use a
973860	975540	voice to text transcription
975540	977240	and then run it through social
977240	979040	media filters, which
979040	981180	does not catch context
981180	983180	or behavioral or environmental
983180	984840	factors. Or
985800	987000	three, they try to
987800	989440	use realtime behavioral
989440	991060	tracking, which is
991060	993220	just not scalable and puts the burden
993220	994240	on users.
997080	999200	When you report a violation
999200	1001360	in a platform like
1001360	1003340	Horizon Worlds, it takes one to
1003340	1004980	two minutes of content
1004980	1007140	and gives it to the company.
1007500	1009200	That normally is not
1009200	1011560	enough to determine the context
1011560	1013500	of what's actually happening.
1013560	1015300	And if you watch videos
1015300	1016880	of people who are reporting
1016880	1018740	abuse happening,
1019200	1021240	you often can't tell who is saying what
1021240	1022960	to whom and what's happening.
1022960	1024940	Especially in 60 seconds.
1027080	1028020	Reporting is done
1028020	1029980	differently as well on every different
1029980	1031960	platform. So that's like
1031960	1034060	if we went to
1034060	1036060	Vegas and you call 911,
1036060	1037460	but then if you went to Henderson
1038180	1040440	and you had to send them a fax,
1040440	1041680	and then if you went to Reno
1041680	1043680	and you had to show up in person.
1044240	1045980	When you use public safety
1045980	1047880	services, if you had to call for help
1047880	1049880	in a different way in every
1049880	1051720	different city you went to, you
1051720	1053760	would never use 911.
1053760	1055420	And I think
1055420	1057220	companies are finding that it's the same way
1057220	1059460	if you don't have standardized safety protocols
1059460	1061400	across different XR platforms
1061400	1062740	and XR worlds.
1066410	1068010	Generative AI looks promising
1068010	1069850	as a tool to help, but it does not yet
1069850	1071490	have the discretion or the contextual
1072070	1073610	understanding necessary to help
1073610	1075430	with the nuance that will be required
1076070	1077530	for helping to modify
1077530	1079430	or govern user behavior.
1080110	1081530	Realistically, AI based
1081530	1083490	solutions are several years away from having
1083490	1085510	sufficient sophistication, enough data
1085510	1087410	and adequate training to work in spatial
1087410	1088630	environments.
1091700	1093920	So, overall,
1093920	1095140	the differences between
1095140	1097140	social media and XR
1097140	1098860	are stark.
1099460	1101300	And they emerge from
1101300	1103580	the interactive features of immersive experiences
1103580	1105200	and the neurological impacts
1105200	1106880	that this has on users.
1107080	1109220	And this has resulted in violations of what I
1109220	1111380	called the right to mental privacy.
1111620	1113420	Other scholars in Chile have called this
1113420	1115180	neuro rights.
1115300	1117180	Brain-computer interface scholars
1117180	1119420	are talking about a
1121460	1121780	battle
1121780	1124200	for your brain.
1124560	1127180	There's a new book that came out
1127180	1129040	discussing that.
1129040	1130440	And they're all
1130440	1132540	part and parcel of the same thing
1132540	1135760	where the sanctity of the content of
1135760	1137240	your thoughts and
1137240	1139100	the inferences that can be
1139100	1141380	made about your health
1139420	1141420	and your sexual
1141380	1143540	preferences, your political
1143540	1145680	preferences, your gender, your age,
1145680	1147400	all of these things we can determine from
1147400	1148100	data flows
1149680	1151460	really shouldn't be used
1151460	1153840	to send you advertisements.
1154980	1155500	If so,
1155500	1157360	I think it's actually a failure of the
1157360	1158400	imagination.
1165340	1167080	I think that's a good
1167080	1168380	way to turn it to us.
1169940	1170780	And one
1170780	1172620	of the things that really
1173440	1174620	strikes my mind is
1174620	1176380	it's not just
1176380	1178420	the gaming experience.
1178420	1180680	This is moving beyond
1180680	1182980	so many different aspects
1182980	1184600	and potential.
1184600	1186280	So as Britton was talking about,
1186280	1188320	moving into the healthcare environment,
1188320	1189720	that your healthcare
1189720	1191920	will be delivered potentially
1192430	1194380	through this data
1194380	1195620	that is being collected
1196410	1199000	from your physical reactions.
1199000	1200540	So the questions
1200540	1202500	start to become, is this data
1202500	1204500	painting a correct picture?
1204500	1206300	Is this the full picture? Is it
1206300	1209140	picking up on all the nuances?
1209440	1210480	And who's getting access
1210480	1212580	to that? And there's
1212760	1214220	a big difference between
1214220	1216700	what I may post on
1216700	1218680	social media at 2am
1218680	1220500	when I am fired up about
1220500	1222580	something and what I would actually
1222580	1224520	say or do.
1224520	1226180	Or you think of a lot of the,
1226180	1227980	as we saw during
1229140	1230420	remote environments
1230420	1232400	where people were having job
1232400	1234540	interviews over Zoom
1234540	1235500	or other
1237040	1238740	conference calling.
1238800	1240420	And a lot of it dealt
1240420	1242240	with, there were companies, and
1242240	1244540	some states have started taking a harder
1244540	1246320	look at this, where they were creating
1246320	1248360	biases. That if you didn't make
1248360	1250300	the proper amount of eye contact,
1250300	1252360	or if it looked like you were distracted
1252360	1254200	during, because
1254200	1255960	the sensors on the screen
1255960	1258620	weren't picking this up properly,
1258620	1260400	that suddenly you were
1260400	1262160	getting penalized
1262160	1264260	during your job interview
1264260	1266200	because you weren't meeting
1266200	1268460	this set standard.
1268460	1270640	And so now what happens when we take
1270640	1272660	that and all of those
1272660	1273340	AI
1274520	1276040	data sets that are
1276420	1278420	expecting certain behaviors, they're not
1278420	1279360	expecting children
1280240	1282480	to start drawing inappropriate
1282480	1284280	things, which
1284280	1287040	anyone who's been in an elementary school,
1287040	1289060	you should have expected it.
1289060	1291520	And you should have expected fart jokes,
1291520	1293040	you should have expected
1293040	1294980	lots of poop,
1294980	1297840	and all those different things.
1297840	1299300	But what happens
1299300	1301400	when we're trying to create
1302160	1303260	protections,
1303260	1304800	and we're trying to do all this
1304800	1307060	with a bias towards, like, we're already
1307060	1309040	starting so far behind.
1309040	1310960	And now we're taking
1310960	1313160	all these other little cues of
1313160	1315360	how does the VR
1315860	1317720	headset, like how does that experience
1317720	1319860	know that it wasn't because a
1320440	1321580	spider crawled across
1321580	1323580	your foot or someone else in the
1323580	1325360	room thought it would be hilarious
1326040	1328440	to tap you on the shoulder
1328440	1330080	as you were playing the game
1330080	1331620	or, or,
1331620	1334100	I'm not saying this has ever happened, I've
1334100	1336500	definitely never done this to someone
1336500	1338060	playing, but what happens if you
1338060	1339700	move furniture while they're
1339700	1341540	in the middle of and suddenly they're
1341540	1342960	tripping over something because it was really
1342960	1344820	freaking funny and
1344820	1346640	it serves them right
1346640	1348800	but suddenly that reaction
1348800	1350920	is getting captured
1350920	1352940	and how do you know that
1352940	1355680	I haven't logged in to Britton's account?
1356000	1356840	That I'm not
1356840	1359000	playing as her and should I
1359000	1360760	be so lucky to
1360760	1362960	have all the sensors pick up
1362960	1364980	and be like, hmm, you have the same
1364980	1367000	facial structure, you have all this
1367000	1368740	and I'm like, yes I do.
1368820	1370980	Yes, that's right.
1370980	1373020	Or, you're just off today
1373020	1374800	or heavens forbid, I think of
1374800	1376820	all the times I've tried to log in
1377500	1378940	to my iPhone
1378940	1380840	and it's like facial recognition, not
1380840	1382700	recognition, I'm like, wait, what?
1383540	1384960	No, oh my god,
1384960	1386980	are you telling me I didn't get enough sleep last night
1386980	1388780	and suddenly, or
1388780	1391500	oh, god, is my resting bitch face back?
1391500	1393160	Like, oh, I didn't mean to do that.
1393160	1395020	So how do regulators and how
1395020	1397120	do you work all those different
1397120	1399000	aspects in? Because we're already
1399000	1400340	playing catch up.
1406240	1408000	One of the things that
1408000	1409620	I like to do when I
1409620	1411500	talk to regulators is
1411500	1414060	tell them the type
1414060	1415560	and quality of information
1415560	1417860	you can get about a person from these
1417860	1419660	data flows. And I realize I didn't
1419660	1421500	mention that before.
1421560	1423760	There was a study that came out last week
1423760	1426220	this is the absolute newest information
1426220	1427240	from UC Berkeley
1427920	1429120	and the study
1430280	1432020	said that a machine learning
1432020	1433960	model was able to accurately
1433960	1436600	predict the height, weight, age,
1436600	1437480	marital status
1438100	1441840	of a user in
1441840	1444220	XR by the way they moved.
1447770	1448010	So
1448010	1450190	I know earlier I said that it doesn't
1450190	1452090	go into the realm of personal identifying
1452090	1454070	information, but initial studies
1454070	1456130	are starting to show that
1456130	1457370	you can actually tell somebody's
1458150	1460170	foot size by the way
1460170	1461130	they move.
1461130	1463050	The lab at
1463050	1464370	Berkeley and the lab at Stanford
1464850	1466510	have had some
1467130	1468890	similar studies and Berkeley
1468890	1470830	came out with a study that
1470830	1471170	said
1471170	1474750	there was 100 seconds
1474750	1476890	of recorded data you could
1476890	1478870	uniquely identify a person
1478870	1480830	in XR with 94%
1480830	1481530	accuracy
1481530	1484650	from a data set of 50,000
1484650	1485810	people.
1486950	1488610	Stanford did this earlier
1488610	1490730	but the data set was
1491170	1492710	a couple thousand people, it wasn't
1492710	1494830	statistically significant.
1495350	1497590	50,000 people.
1497590	1498570	That might be
1498570	1500870	you in the Taylor Swift crowd
1500870	1502610	by the way that you tilt
1502610	1504730	your head and point can be uniquely
1504730	1507390	identified in an immersive environment.
1508310	1509030	The type
1509030	1510830	of medical
1510830	1512710	and sort of human rights
1512710	1514850	oriented information that I'm concerned about
1514850	1516850	really is
1516850	1518350	tied into eye tracking.
1519270	1520890	So the type of information you can get
1520890	1522710	by somebody's pupil reaction
1523550	1524770	you can tell who
1524770	1526050	somebody is sexually attracted to
1527010	1528830	you can tell whether or not somebody is likely
1528830	1529690	to be telling the truth
1530430	1532650	and you can also tell whether or not somebody
1532650	1535110	shows preclinical signs of certain
1535110	1536390	medical ailments
1536390	1538650	like schizophrenia, ADHD
1538650	1540910	Parkinson's and Huntington's.
1541010	1542370	These are preclinical signs
1542370	1544930	these are things you may not know about yourself
1544930	1546710	yet, your doctor may not
1546710	1547810	know about you yet
1547810	1551130	and a company
1551130	1552790	if they were to
1552790	1554870	parse the data set would have
1554870	1556950	access to that information.
1557630	1559050	This is what I mean
1559050	1561290	by the right to mental privacy.
1562150	1562810	It's not
1562810	1564670	just the content
1564670	1566630	of your thoughts, it's the
1566630	1568830	implications that can be made about
1568830	1570890	who you uniquely are
1570890	1573030	your medical conditions
1573030	1575310	and the type of behaviors that you may
1575310	1577830	choose to engage in or not.
1579950	1581130	One of the things
1581130	1583130	we're seeing is the states are starting
1583130	1584550	to try to play catch up
1584550	1587310	because, keeping in mind
1587310	1588790	data brokers
1588790	1590990	are taking this information and
1590990	1593350	creating these pictures
1593350	1594850	and they're
1594850	1596670	selling them. And
1596670	1599410	the regulations really are not
1599410	1601370	keeping up
1601370	1603230	with who should have access
1603230	1606090	to this and how do we protect it.
1606290	1607710	And it's also
1607710	1609270	what happens if
1609270	1611410	it gets wrong.
1611630	1613270	That one
1613270	1615450	out of 50,000
1615450	1617370	it's wrong.
1617610	1619410	But yet this is now
1619410	1621410	flagged in a file and it's
1621410	1623390	you know as a child you're told this will go
1623390	1625230	down in your permanent record.
1625850	1627530	Well how do we fix that
1627530	1629430	permanent record when it's our
1629430	1631530	biometric and it's
1631530	1633370	our mental information
1633370	1635210	that, oh wait, you know,
1635210	1637650	objection your honor. I was just
1637650	1639810	having a bad day and that wasn't me.
1640450	1640970	Or
1640970	1642890	the, you know, it
1643350	1645470	that was someone else and that
1645470	1647510	piece of information. So it's not the
1647510	1649170	entire picture that gets off.
1649170	1651530	It's just one piece. How do you fix
1651530	1652350	that?
1656970	1658810	I don't have a good answer for that.
1658810	1660430	I wish I did.
1660430	1662190	Because right now
1662190	1664250	these type of data sets aren't
1664250	1665970	auditable by the public.
1667070	1668410	In certain jurisdictions
1668410	1670330	people are proposing a right to have
1670330	1672430	all your user information about you
1672430	1674510	available, but these are
1675610	1676590	profiles made
1676590	1678090	to sell you something.
1678390	1680110	It's not an essential part of your
1680110	1682130	user account and things.
1682870	1684530	Another thing I tell regulators
1684530	1686430	is that you
1686430	1688390	can't outsmart the machine.
1688390	1690130	Because a lot of them think that
1691310	1691830	somebody
1692490	1694650	advertising something to me, that's obvious.
1694650	1696330	That's not
1696330	1698630	I'm not going to fall for that.
1698890	1700290	In XR environments you can actually
1700290	1701870	switch out elements
1702390	1704910	in the environment at blink rate.
1707550	1708070	It's
1708070	1710590	been done in experimental context
1710590	1712250	but there's nothing saying that you could
1712250	1714110	have somebody with a can of soda
1714110	1716330	and switching out Coke and
1716330	1717850	Pepsi and
1717850	1720310	tracking somebody's more
1720310	1723130	positive reaction to one versus the other.
1723510	1724350	It's kind of like
1724350	1726330	that social psychology experiment where
1726330	1728490	you're told to count the number of
1728490	1730390	basketball passes and so you
1730390	1732830	miss the gorilla walking in the background.
1733930	1734390	It's
1734550	1736070	a similar type of a dynamic
1736070	1738470	where people think they're smarter
1738470	1740370	than the system, but the system
1740370	1742250	works faster than you can
1742250	1744350	articulate your preferences or
1744350	1746450	requests cognitively.
1747670	1748010	And that's
1748890	1751010	a really kind of funky, scary thing.
1751110	1752330	If you watch
1753190	1753870	sports
1753870	1756430	and Liz is here and I'm back in
1756430	1758410	California, Liz
1758410	1760590	and I will actually see different ads
1760590	1762110	behind the players.
1762950	1764470	Augmented advertising is
1764470	1766350	already a thing and
1766350	1767610	it's based on your
1768530	1770190	zip code.
1771050	1772470	And somebody in the stadium
1772470	1774410	will see even different
1774410	1776450	ads than Liz or I would see
1776450	1778230	on different network channels.
1778250	1780190	They sell this space.
1780410	1782530	Advertisement in an immersive
1782530	1783950	context is more about
1785090	1786970	targeting and personalization
1786970	1788450	than it is
1788450	1790790	in an offline context. And that makes sense,
1790790	1791890	but the capacity
1792470	1794590	and the ability to really
1794590	1795790	target it to people
1796550	1798610	makes XR the most persuasive
1798610	1800510	medium that we've
1800510	1801710	developed so far.
1803010	1804490	And one of the things that
1804490	1806190	we saw and we keep going back to is
1807170	1808690	the data sets.
1808690	1810390	What is it getting based on? Where
1810390	1812370	are those data sets getting built from
1812370	1814330	that are creating some of these
1814330	1816010	models and
1816710	1818250	being able to audit it
1818250	1820610	and being able to audit and get to your
1820610	1822430	information and make sure, like, hey,
1822430	1825010	is this the picture that I see?
1825210	1826390	And basing it
1826390	1828530	on a bias towards,
1828530	1830430	well, English
1830430	1832790	speaking, US
1832790	1834510	based things, when you're
1834510	1836670	talking about an immersive experience
1836670	1838450	that is intended to
1838450	1840610	open up the world.
1840610	1842470	Like, the idea is that you
1842470	1844430	were playing and you were experiencing
1844430	1846530	things with people
1846530	1848730	in any different physical
1848730	1850610	location. So now
1850610	1852370	how do you build the
1852370	1855010	rules and even train
1855010	1856270	that it's getting it right
1856270	1858390	when not
1858390	1860410	everyone's going to look the same, react the same
1860410	1862530	and different cultures have built
1862530	1863790	in different
1865370	1866490	mannerisms.
1866890	1868590	And as we saw with a lot
1868590	1870390	of the, you know, again,
1870390	1872390	even just trying to unlock your iPhone,
1873050	1874570	heavens forbid, you're
1874950	1876490	a woman of color.
1876490	1878390	Because guess what? All of
1878390	1880330	the modeling that was based
1880330	1882210	on was not you.
1882350	1884450	Not even a little bit. And so they had
1884450	1886850	to go back and say, oops, our bad.
1886910	1888310	Well now, what do we do when we're
1888310	1890470	talking about collecting all this data
1890470	1891930	and it's like, on
1892310	1894270	the reactions of
1894270	1896410	you know, we talk about a lot in
1896410	1898470	the medical world, a lot
1898470	1899430	of the early
1900330	1902470	training was done and it wasn't
1902470	1904530	done based on women's
1904530	1906490	bodies. And it wasn't based
1906490	1909070	on... so it's like, oh, well,
1909070	1910350	you're going to react to this
1910350	1912450	medication in the following manners, or these
1912450	1914390	are the following symptoms. It's like, but
1915930	1917290	no.
1917550	1918710	I am not.
1918710	1920410	So now you've got the VR and
1920410	1922530	it's assuming, heavens forbid,
1922530	1924510	I don't want to be
1925990	1926670	a redhead
1926670	1928330	middle aged woman in
1928330	1930530	my, like, I want to
1930530	1932710	be a kick ass
1932710	1934410	like ninja
1934410	1936430	dragon, whatever the hell
1936430	1938390	I want. So
1938390	1940330	how are we going to start
1940330	1942370	building that and work on regulators when you've got
1942370	1944490	because, again,
1944490	1946670	heavens forbid, heavens forbid
1946670	1947770	that
1948450	1950390	Australia, the US, New Zealand
1950390	1951310	like, insert
1952150	1954290	country, get the upper
1954290	1955790	hand in being
1956330	1958170	the regulator who
1958170	1960330	sets the rules for this.
1960330	1962370	Because, oh no, we can't let
1962370	1964290	Canada do that. I mean, they're just nice
1964290	1965750	and maple syrup.
1966450	1968510	So how do you broach
1968510	1969670	those conversations
1970610	1972570	when you're going all over the world
1972570	1974410	and talking to them and saying, no, no, no,
1974410	1976350	we need to let this country take the lead
1976350	1977990	or play nice with them
1978410	1979950	and not share the data.
1986860	1988200	If y'all notice, I give
1988200	1990460	Britton all the really hard questions
1990460	1992180	because this is what she gets for being
1992180	1994120	so smart and being a professor,
1994120	1995720	shaping young minds
1995720	1998200	that are creating this future. So, sorry.
1998220	2000120	But those aren't young minds, those are like
2000120	2002020	crusty old minds. That's why I'm like,
2002020	2003640	hmm, about this.
2005080	2006100	A lot of people
2006100	2008240	in Europe argue that GDPR
2008240	2009800	should cover it.
2009960	2012000	They argue that Article 1 of GDPR
2012000	2014060	should cover all of the privacy protections
2014060	2016180	for XR and then we're done.
2016460	2018220	I take issue with that.
2018780	2020260	Because these are new types
2020260	2021800	of data flows and
2021800	2024060	they're not really anticipated by
2024060	2026000	the regulatory structure.
2027220	2028340	And they're
2028340	2030020	more akin to medical data
2030020	2031640	to me than they are personal
2031640	2033540	identifying information.
2036800	2038340	The Future on Privacy
2038340	2040520	Forum just did a
2041700	2042980	really optimistic
2042980	2044220	study where
2044220	2046480	they determined that biometric laws
2046480	2048260	are, biometric laws in the US
2048260	2050960	are covered by state based law.
2050960	2052520	And if you have a business
2052520	2054020	that creates XR
2054720	2056240	products, you know this because there are
2056240	2058020	certain states you're really, really worried
2058020	2060020	about selling in and certain states
2060020	2062140	you may choose to not release your product
2062140	2064320	in. And those may be
2064320	2066100	Illinois, Washington, or
2066100	2066940	Texas.
2068200	2069880	There are other states that have privacy
2069880	2072180	laws, not biometric laws, but
2073720	2074240	you
2074240	2076320	think about, sometimes I talk to regulators
2076320	2078180	and I ask them if they've tried
2078180	2079480	Smell-O-Vision.
2080300	2082260	And they look at me like I'm nuts.
2082580	2084200	And I tell them that in the lab
2084200	2086120	and in commercial context now,
2086120	2088600	you can actually go before a
2088600	2090640	campfire in XR and you don't
2090640	2092560	just hear the campfire and you don't just
2092560	2095200	see the campfire, you can actually
2095200	2097060	smell the charcoal
2097060	2098900	and in the lab we put a little
2098900	2101020	toasted marshmallow in there as well just for
2101780	2102500	fun.
2102500	2107860	You can go into a garden and smell flowers.
2107860	2109240	These type of inputs
2109240	2110920	are really not anticipated
2110920	2114080	by law.
2112120	2114120	Also,
2114080	2117280	there was a
2117280	2118080	company that
2119900	2121020	recently released a
2121900	2122460	breath
2122460	2125120	gauge for a meditation app.
2125120	2127320	And my question for them is
2127320	2129140	is that biometric data?
2129140	2131120	It could be body-based data
2131120	2132340	based on some of the states
2132880	2134580	so it would be regulated as such.
2134580	2136880	In other states, it wouldn't be.
2136880	2138980	I think the way that I make regulators
2138980	2140440	understand this is I
2140440	2142120	try to actually bring them into the
2142120	2143460	developer's mindset
2143940	2145400	and bring them into
2147680	2150500	the chair of somebody who has to make a
2150500	2152540	decision about
2152540	2154620	selling and producing these products
2154620	2156560	and doing so in an
2156560	2163830	environment that is ambiguous at best.
2163830	2167070	Some countries are better about it than others.
2167070	2167670	Australia
2167670	2169450	has an e-safety commissioner's office
2169450	2171750	so they're very, very motivated.
2171750	2173350	New Zealand is very concerned about public
2173350	2175310	safety in the online
2175310	2177370	context. France has had
2177790	2179530	a national metaverse strategy.
2179950	2181110	China and
2181110	2182630	Colombia and
2183990	2185270	one other country
2185270	2187210	have started to hold trials in the
2187210	2189350	metaverse. Don't do that
2189350	2191730	by the way. That's not a good idea.
2192990	2195350	Countries are starting to engage in little
2195350	2196930	ways. I think
2196930	2198330	Bahamas had
2200030	2201170	an embassy
2201170	2203510	in the metaverse.
2204830	2205610	I think
2205610	2207370	the way to actually get to these countries is
2207370	2209490	to talk to
2209490	2211550	them more about their economies
2211550	2213310	and corporate environments and that
2213310	2215570	should get them more motivated rather
2215570	2217390	than talking about digital diplomacy
2217390	2218510	and
2219630	2221570	kind of a new
2221570	2223130	sovereign space or other
2223130	2225490	type of nonsense things.
2226330	2227450	When you think
2227450	2229730	about how even in the U.S.
2229930	2230530	a lot of our
2231790	2233970	laws that touch on
2233970	2234910	things that we deal with
2234910	2238050	day to day, the DMCA,
2238050	2239850	CFA, a lot of them came
2239850	2241930	because members of Congress
2241930	2243830	for example were afraid that their
2243830	2245890	blockbuster rental history
2245890	2248470	was going to be published.
2248990	2249950	Oh yeah.
2249950	2251670	Or they saw a movie
2251670	2254470	and they said, you know what?
2254470	2256490	War games. Huh.
2256530	2257510	Huh.
2257510	2259470	So I love
2259470	2261310	that idea of putting it in that
2261310	2263330	personal context because
2263330	2265670	unfortunately that does
2265670	2266950	ground it for them
2267370	2269310	and show them because so often we
2269310	2271350	spend time chasing the technology
2271350	2273630	like oh we have to react
2273630	2275150	to this when it's like no
2275150	2277290	let's look at what it's
2277290	2279610	actually doing. What is it doing at its core?
2279610	2281750	What is the actual concern?
2281770	2283750	And then how do we build from there?
2283750	2285510	So I love that you're
2285510	2287510	getting them that hands on but how do y'all
2287510	2289590	scale that kind
2289590	2291350	of advocacy or how do we
2291350	2293450	as a community
2293450	2295630	of users who are concerned
2295630	2297170	about or builders
2297170	2299770	how do we take that
2299770	2301790	idea and that approach
2301790	2303070	and scale it
2303070	2304070	beyond
2305270	2307110	and bring it down to that
2307110	2309470	local level or that regional level
2309470	2310750	of hey
2311370	2313470	do this, see this, experience this and know
2313470	2315670	that this is going to impact what you're doing.
2316490	2317550	I think you don't
2317550	2319610	use law. I want
2319610	2321210	to call out someone in the front row
2321210	2323450	Caroline Cinders who's
2323450	2325690	research about how to use effective
2325690	2327570	design to create healthy online
2327570	2329510	communities and environments is
2329510	2330810	awesome.
2332170	2333270	Yay!
2333910	2335290	So looking at
2335290	2337410	design based principles,
2337410	2339290	looking at social psychology
2339290	2341590	about how we create cultures
2342570	2343590	some of my
2343590	2345690	very boring academic work focuses
2345690	2347850	on the theory
2347850	2349590	of code as law and how
2349590	2351910	that has ignored norms.
2352210	2353650	And norms are the way that we create
2353650	2355830	culture and we self regulate our own
2355830	2357650	online environments.
2357650	2359370	And try not to place
2359370	2361250	the full burden on
2362210	2363390	users but
2363390	2365490	distribute it collectively amongst
2365770	2367150	a whole community.
2368170	2369630	It seems
2369630	2371530	like a cop out answer because I'm a lawyer.
2372830	2373790	But this
2373790	2375510	isn't a legal problem I think.
2375510	2377710	It's more of a
2377710	2380170	social problem and an issue
2380170	2381810	of science not catching up
2381810	2385510	to politics.
2385510	2385910	Well and I
2385910	2389050	think that also putting it on the users
2390090	2390930	overlooks
2390930	2392370	the builders, the makers.
2392370	2394650	Like if we are intentional
2394650	2395990	from the design aspect
2395990	2397770	and when I say we, I mean the people
2397770	2399250	at the companies who are building it
2399570	2401770	from product to engineering
2401770	2403670	to all that understanding and
2403670	2405410	seeing that big picture
2405990	2407670	and knowing when to call it
2407670	2409490	oh this is going to impact this.
2409750	2411770	Let's be intentional. Let's bring them into the
2411770	2413750	conversation. So as
2414330	2415690	a DEF CON community
2415690	2417950	I think that's one of the easiest ways
2417950	2419710	to start. I was like
2419710	2420990	start the change.
2421210	2423490	And I think we can also be really
2423490	2425990	intentional about how we design
2425990	2427950	these products and who
2427950	2429990	we are accommodating
2429990	2431910	when we design them.
2432730	2433910	How many people
2433910	2436690	here are familiar with simulator sickness?
2436850	2438270	Probably a bunch.
2438310	2439770	If you're not, it's
2439770	2442090	if you've gone into a VR world and you
2442090	2443990	feel nauseous and sick to your
2443990	2445710	stomach and really disoriented
2445710	2448370	that's simulator sickness.
2448890	2450310	In the earlier
2450310	2452050	iterations of headsets
2452770	2454290	more women than men
2454290	2456610	reported simulator sickness.
2456610	2458330	People started to write about this.
2458330	2460090	Some said it was just my delicate
2460090	2461190	female constitution
2461830	2464030	and the fact that I may
2464950	2466590	produce babies one day
2466590	2468310	or the fact that I
2468310	2470290	don't play first person shooter
2470290	2472510	games. It's all bullshit.
2472910	2474270	What actually was happening
2474270	2476030	was that the headset was designed
2476030	2478310	for an average male form
2478310	2480150	between 5'10 and
2480150	2482410	6 feet tall. There is one
2482410	2484330	measurement that is essential to preventing
2484330	2486450	you getting simulator sickness.
2486450	2488890	It's the interpupillary distance.
2488890	2489870	The distance between
2490410	2492370	your eyes. If you go to
2492370	2494410	the optometrist, they measure this to put you
2494410	2496470	in a pair of prescription glasses.
2496890	2498270	The interpupillary
2498270	2500670	distance on early VR headsets
2500670	2502270	was not adjustable.
2502570	2504610	So it was akin to putting 51%
2504610	2506490	of the population in the wrong
2506490	2508050	pair of glasses.
2508710	2510590	No wonder women were reporting
2510590	2512990	sickness and discomfort.
2513130	2514610	And there were studies following that
2514610	2516730	by Jessica Outlaw that actually showed
2516730	2518630	people that women weren't returning to
2518630	2521070	VR environments because they felt uncomfortable.
2521830	2522770	There were similar
2522770	2524270	studies that came out.
2524270	2526930	One from the MIT Media Lab
2526930	2528930	where there was a researcher doing her
2528930	2530590	master's thesis work in
2530590	2532790	Nairobi. She took Oculus Go
2532790	2534390	to try to put it on
2535230	2536430	Kenyan entrepreneurs
2536430	2538730	and she found that with the women
2538730	2540430	50% of the time
2540430	2542370	the strap snapped
2542370	2545210	when she tried to fit it over their hair.
2545710	2547130	50%
2547130	2548570	of the time.
2548670	2550730	She ended up deviating from
2550730	2552710	her master's thesis to design an
2552710	2555190	adjustable fucking strap.
2556050	2556810	Easy
2556810	2559010	fix if you think about the wide
2559010	2560850	beautiful variety of people
2560850	2562750	who you want to be
2562750	2564790	your customers. It's smart
2564790	2566770	to think about the wide diversity
2566770	2569090	of body shapes and types.
2569970	2571130	Disabled populations
2571130	2572890	are some of the earliest adopters
2572890	2574510	of XR hardware.
2574830	2576730	And they are the last people that many
2576730	2578790	companies think about when they design
2578790	2581190	these products and services.
2581530	2582770	Oculus didn't change
2582770	2584790	their vantage point to make
2584790	2586630	it adjustable whether or not I was
2586630	2588830	standing or sitting until
2588830	2590350	update 31.
2591130	2592670	And it came on the heels of a Scientific
2592670	2594670	American article that showed
2594670	2596370	that somebody had to modify an
2596370	2598750	Xbox controller to use a rock
2598750	2600670	climbing game. They were
2600830	2602590	a user with muscular dystrophy and so
2602590	2604650	they really wanted to know what rock climbing
2604650	2605830	felt like.
2606490	2608370	And they had to modify it themselves
2608370	2610910	because the game wouldn't accommodate them.
2612970	2614590	This is something that we
2614590	2617130	can do as a community.
2618370	2618670	So
2618670	2620410	when you're doing your
2620410	2623070	prototype testing or your
2623070	2624570	platform building or
2624570	2627110	your experiential
2627110	2629050	design, think
2629050	2631090	about all of the people you want
2631090	2633070	in your world. And then go
2633070	2633910	bigger.
2634930	2637370	Yeah. And I think bringing it back
2637370	2639070	to what
2639070	2641070	Britton was talking about before is like that full
2641070	2642950	stack of an
2642950	2644890	understanding it's not
2645670	2647130	a level playing field. And by that
2647130	2648790	I mean it's not a flat
2648790	2650770	screen. And so
2650770	2652450	as you're building and as you're
2652450	2654670	designing, taking all of those different
2654670	2656830	aspects and
2656830	2658810	also incorporating it
2658810	2661170	into the conversations
2661170	2663130	on the biometric data.
2663130	2664810	What exactly? So we're not
2664810	2667090	using those same definitions.
2667550	2669390	It's not the same game.
2669390	2670910	And being, again,
2670910	2672490	intentional in how you're
2672490	2674890	building, how you're designing, but also how you're
2674890	2676610	approaching it. And
2676610	2678830	be it with the conversations you're having
2678830	2680970	with regulars as you're watching
2680970	2682770	that. I think really
2682770	2684910	just kind of bringing it all. This is
2684910	2686890	not, you know, this is
2686890	2688850	an immersive environment. It is the
2688850	2691250	entire all encompassing environment.
2691670	2692790	And there's some best
2692790	2694950	practices that you can follow as
2694950	2696130	developers
2696740	2699010	or even as security experts.
2699090	2700310	A best practice
2700830	2702910	for XR is to actually
2702910	2705330	keep the data on device.
2706230	2706630	That
2706630	2708470	means that you have to get a warrant to
2708470	2709890	access the data.
2710550	2712470	The problem with that is
2712470	2714810	it really impedes
2714810	2716490	your processing speed. It's
2716490	2718590	very, very expensive.
2718710	2720590	It kind of drains your battery.
2721310	2722530	So the other thing you can do
2722530	2724710	is when you're looking to third party
2725550	2727450	cloud providers,
2727450	2728690	you can be really deliberate
2728690	2730610	about asking them about their data
2730610	2732840	retention, storage,
2733970	2735130	disposal,
2735130	2736710	all of their data policies
2736710	2738890	because what your third party data provider
2738890	2739410	does
2740870	2743230	is basically what you do.
2743310	2745850	And being a savvy customer.
2745970	2746890	So pushing back
2746890	2749010	if there aren't privacy protective services
2749010	2750750	like you feel meets the threshold
2750750	2751570	that you want
2752830	2755110	to protect your users.
2756090	2756810	And that's
2756810	2759410	excellent because the change starts with calling it out.
2759690	2760970	And bringing it
2760970	2762770	to the attention of and paying attention to
2762770	2764690	because you also think of all the different users
2765290	2766750	and whether they're going
2766750	2767970	to understand
2769730	2771050	a 12 year old child
2771050	2773090	is not going to care
2774110	2774550	about
2774550	2776370	all their different things, but it's like
2776370	2778510	all they know is there's a lag time.
2778810	2780330	And they're annoyed. It's like, no, let's
2780330	2782950	make sure that we are paying attention,
2782950	2784590	that we are bringing it in, and that we are
2784590	2786570	reporting it, raising it.
2786570	2788450	It's kind of the name and shame.
2788970	2790630	Hey, this is a great game that you've
2790630	2792270	created for children or geared
2792270	2794310	towards this market.
2794390	2796430	Now you're creating their permanent record
2796430	2797810	for their school.
2798110	2800630	The equivalent of that example
2800630	2802630	that's going to follow them forever
2802630	2804110	and is going to be based on
2804110	2806230	AI modeling and
2806230	2808250	data points that are not
2808250	2810490	necessarily accurate. I mean, they're going to be based
2810490	2812690	on, if we're looking at the early sets,
2813670	2814650	the 5,
2814650	2816470	11, 5, 10
2816470	2817310	male
2819670	2820270	interactions
2820270	2821490	and all of that
2821490	2822770	different thing.
2822770	2824890	If you had to think of
2824890	2827450	one way to bring it
2827450	2829350	home, of how you're
2829350	2831310	dreaming of electric sheep,
2831910	2833370	what are some good takeaways
2833370	2835510	other than, I mean, I love what you were saying
2835510	2837610	with the proactive
2837610	2838970	steps.
2839030	2841270	What's a good nugget?
2841570	2843230	What's the quote?
2845790	2847250	It's the hangover spot
2847250	2849690	on Sunday morning. I don't know if I have a quote.
2860350	2861450	Someone asked what is
2861450	2863530	inspiring in this space, which is
2863630	2865190	a great question to end on because
2865890	2867370	it's all doom and gloom
2867370	2869010	when I go to the dinner party.
2869750	2871290	XR is magic.
2872510	2873830	It's magic.
2874570	2875590	It makes, it can
2876810	2877650	it can make
2877650	2879850	you feel
2879850	2881830	like you are in a different body.
2882070	2883410	You start to embody
2883410	2885810	the characteristics of your avatar
2885810	2887470	in a social environment
2887470	2889530	within two minutes.
2889990	2891390	So if I make myself an
2891390	2893590	eight foot tall dragon, I'm going to dominate
2893590	2895690	that space in two minutes.
2896450	2897690	It takes two and a half
2897690	2899610	minutes for me to put you in
2899610	2901550	my lab context and give you a third
2901550	2903570	arm and have you pop three balloons
2903570	2905230	at the same time.
2906570	2907110	Neuroplasticity
2907110	2909770	is magic. XR is a
2909770	2912550	means to access your neuroplasticity,
2912550	2914370	your imagination is
2914370	2916510	really the limit. And we've never had
2916630	2918710	a technology like this before.
2918930	2921210	So let's not fuck it up.
2921510	2922890	That's my quote.
2922970	2924330	I like that.
2924330	2926530	Thank you so much and thank you to
2926530	2928450	the XR village for
2929770	2930410	bringing
2930410	2933310	this topic, inviting us.
2933410	2934630	Please go check it out
2934630	2936450	if you haven't already. I'm
2936450	2938650	Lawyer Liz on Twitter. Please feel
2938650	2940710	free to engage.
2940710	2942750	I'm Britton Heller. You can
2942750	2944930	find me as Britton Heller on
2944930	2946990	Twitter or on LinkedIn.
2948170	2948910	Thank you
2948910	2949930	so much.