Player is loading...
(Keynote Talk) Biometrics in the era of AI: From utopia to dystopia?
Dr. Catherine Jasserand, KU Leuven (Belgium), Marie Skłodowska-Curie fellow at Biometric Law Lab
Wednesday, Oct. 11, 2023 · 11:06 a.m. · 35m 16s
In her talk, Catherine Jasserand will explain how biometric technologies are perceived and the reasons why it is very difficult to agree on their role in a democratic society. She will also discuss the notion of technology neutrality and explain the current debate on the regulation of biometric AI systems in Europe.
Embed
Transcriptions
Note: this content has been automatically generated.
00:00:02
ah hello remote come back from lunch break or group you you're
00:00:09
and i'm ready to continue our supporters year extension yeah i your and ethics
00:00:15
ah or speakers recession and our first speaker it is doctor catherine just see how
00:00:22
from my university you're a lowering your uh autograph during the user mary's good oscar or
00:00:30
you know or a parametric low lap so please go through thinking remote carmen and yours yours
00:00:40
ah so thank you very much and probably kind indeed annotation by then there is the best you know
00:00:46
uh it's it's nice for me to be uh in the middle of
00:00:50
of exponents because i'm not a by a metric expert but a legal experts
00:00:54
and uh yeah when uh i was thinking about uh which topic to advise of course that we talk
00:01:00
about uh de uh have a proposal for the
00:01:04
a a yeah act and explain where the negotiation are
00:01:08
but i was thinking that uh um how do we can see biometric technology as we know societies and that's
00:01:14
why i choose to do it and the title it talks yeah for me to get to the stock yep
00:01:19
uh it's it's neither but but i decided to ditch is this title because it's very important
00:01:24
that we understand why there is maybe so much resistance and white still what if you caught
00:01:29
to find a new g. t. makes use of the technologies uh in our a democracy and and society
00:01:36
so um it's it's about giving you some context um so the overview of that so
00:01:42
i'm going to talk a bit about it took and is talking on maybe you know it's
00:01:46
uh it's it's not this is only a only links to
00:01:50
about technologies but i would look at a painting to technologies
00:01:54
did i want to address a topic which i found very important um for researchers
00:02:00
and for the industry is about technology new twenty t. and what it means
00:02:05
and i will uh i would just point out to some research by because
00:02:10
of errors and it is easy to uh oh which trench uh these technology neutrality
00:02:16
and then i will discuss about the deregulation of for me for a
00:02:21
i so the future a acts and explain uh where the issues are currently
00:02:27
and the channels that we have to agree on on this on this
00:02:30
topic so uh i've i've uh i've worked mainly on fisher recognition technologies
00:02:36
i'm not going to rise all the biometric technologies that are covered by the feature octave should enjoy don't act
00:02:43
meaning be emotional recognition and the barometric categorisation systems
00:02:47
but i will just briefly talk about them as well
00:02:51
so concerning um the top yeah and just of jobs
00:02:56
so um i assume that an interview uh no science fiction if
00:03:01
if you work in the field uh you might know since diction
00:03:04
and a new as well that we always talk about these two payout does to plan a won't
00:03:10
uh that's ferrous days it took actually it was for us this idea would
00:03:13
be to clearly would have any day and society uh well no countries necessary
00:03:17
because people will behave in a it the other way and uh we have
00:03:22
any their form of government as we could be society doesn't exist of course
00:03:27
and then uh with the science fiction only tonight you has when uh
00:03:31
uh but but it's it's previously uh it it was a much be
00:03:34
first inspection also the term this type that came out like being the
00:03:40
words from society it's it's a nightmare to be to even such a society
00:03:44
uh what is interesting is that these two tams it would have been a calling them by
00:03:49
yeah i'm pretty sure uh uh by uh to mess more far directly to
00:03:54
chow and uh by if you're so far produced reserve fell for the stock yep
00:03:59
uh and they are critical representation of so say t. i. as well and you try
00:04:03
to i'm sure that you only know uh in nineteen eighty four uh by george orwell
00:04:08
it's not so much about the technology but it's about the government trying to control every aspect
00:04:12
of people's life throw technologies uh technology are helping uh in two men saying b. b. over there
00:04:19
uh uh i'm sure as well that you heard of some of you of be an immense they'll uh by
00:04:25
my guide attitudes or uh maybe more recently because that's
00:04:28
good that you can see a everywhere climb to send
00:04:32
um kansas it's it's an interesting book because it's about a a human you
00:04:37
mentally into robots that because um that becomes a friends uh two is sick child
00:04:42
and it's also makes a job all um that you can you can read uh if yeah if you're interested
00:04:49
uh so all these books our presentation of how technologies as well we'll
00:04:54
yeah we'll be part of our society might be part of our society
00:04:58
but i think the rally teasing between uh in its and then you have concept let that stop you out or put up yeah
00:05:05
i don't know if you heard about these uh concepts uh has to complete the job so to stop yeah
00:05:11
it's it's a concept that in my guides out those whenever i put him up the we because she said that
00:05:17
we have elements of the stock in our world uh it's not a two p.
00:05:20
of course that we are currently even but we have agreements of of the distant yeah
00:05:25
and it's also something that uh who haven't janine
00:05:28
chooses sociologist in in a african american studies nikki states
00:05:34
and she she'll actually that these technologies a. i. technologies not necessarily biometric technologies
00:05:40
they are also uh used uh um to um to frighten
00:05:44
discriminates if people and the and minorities of in our reading about
00:05:49
people so that's that's very interesting to have these uh these drastic t. will you also happen to p. r. it's it's like
00:05:56
you would consider that society is not static and there is always hope
00:05:59
for better so you have uh the hope to have a business eighty
00:06:04
but then i i wanted to give you what i found uh in science fiction and uh
00:06:08
but that's in the real world about uh these relation to uh it took a distant yep
00:06:14
i don't know if you uh i'm not sure you can read
00:06:16
do you know these biometric marrow it's it's an experiment setup that by
00:06:22
and at least uh from us right now um and she instead it's uh it at different uh location
00:06:29
like a museum of amsterdam so i i went there i was with my son it was very interesting
00:06:34
because you've got an r. r. and the mar was kind of phase so it's it's it's can analyses
00:06:40
and according to this kind they will tell you if you are
00:06:43
self confidence and the percentage of sin convenience and ease of use senior
00:06:48
and then that idea it's the project an image of one hundred percent
00:06:52
of each of these characteristics or self confidence uh uh relax a beauty
00:06:58
and the face that you say you can select a you can check and if you do online you too
00:07:04
it's unlikely interface so this is very interesting to see how a high activity
00:07:08
interprets of course is what we f. a. not is there is that right
00:07:12
they can interpret beauty um so i thought that was
00:07:16
that was a different perspective the on this the maybe uh
00:07:21
imaginary societies and how we could use of a matrix but then you
00:07:25
have another one and this is back in your right beside uh uh
00:07:29
we we could side i think most obese it is that of a
00:07:32
black morrow because most of them are the relates also to biometric technologies
00:07:37
this one is is quite intriguing because it's about uh making
00:07:42
alive uh that a depressant again so by creating a digital avatar
00:07:48
but the thing is that the person looks like speaks like but it of course
00:07:52
it's notes because the brain has not been scanned so it's the behaviour which is can
00:07:57
and the choice let's issues because some people don't accept uh to
00:08:00
lose of course the abilities and what if we could do that
00:08:04
why don't we it because he needs as well that we should put to this kind of experiments i'm
00:08:09
not talking about he got it because what should be uh the the the the moral uh boundaries as well
00:08:16
how should we goes when it we shouldn't go ahead and that's the whole question as well with biometric technology because
00:08:22
they can provide so many means uh to achieve
00:08:26
whatever we want or in the future even even more
00:08:30
so then i wanted to to go back to the one that we have a currently
00:08:35
and it's it's like two sides of a point and and uh
00:08:38
i have very simplistic a representation of that's for sure once biometric technologies
00:08:44
i was thinking of this me and maybe uh you know at that particular to do this me a
00:08:51
box they just touch your fingerprints they said convenience security
00:08:56
yeah they don't inform us like but that's another issue
00:08:58
but for them this is either situation you don't have to use in email anymore payment means well
00:09:05
because you uh easily uh button to fire yourself with your fingerprints and you've got
00:09:10
a carte blanche to it and then you also have these uh uh but your fingerprint
00:09:17
cool it's a unleash a lot of uh their applications you can play with a few complaints
00:09:23
you can unlock your fallen uh you can do a
00:09:26
your shopping online you can have access to all the services
00:09:29
so for me these are seen as convenience and it's it's
00:09:33
i think it's easy to illustrate that with verification slash with
00:09:37
authentication uh uh press this is our uh because these uh
00:09:42
then base will uh have less implications in terms of fundamental rights
00:09:47
but then the other side so it's a selection not uh
00:09:50
it it's that's a completely uh uh impartial but what i'm thinking
00:09:55
this one really really trouble me it's about a lie detector and i'm sure that you heard
00:10:01
about that so we could discuss with on presenting a biometric technology per se because they don't always
00:10:08
they are not willing to buy much recognition but stay the a a process by which the characteristics
00:10:14
and this uh i thought of course for it's a project that was the finance by the european commission to try
00:10:20
to um to needs um in the the flow over
00:10:24
of new grants in the u. but the problem of that
00:10:28
is um is is the assumption that you can read a motion
00:10:34
on the face of people when they cross the borders whereas when people cross
00:10:39
the border especially the heart in a stressful situation began should different type of emotions
00:10:44
and we know that emotions are the expression of emotion is not us only deepens as little function to go
00:10:50
it depends on many things so we are funny actually this emotion recognition are really problematic
00:10:56
and we find them in d. f. h. artificial intelligence act so the the this this was a very bad the
00:11:02
project that a receiver that of it produces not answer the
00:11:06
whereas some um trial in place that i've been a start
00:11:11
uh they had planned to try in greece hungary unless yeah and uh after lots of
00:11:15
the uh there was a beagle try and and the and the all these experiments have stopped
00:11:21
but i was thinking of of peace one you might see that it was last week or two weeks ago
00:11:27
uh it did make a concerning uh tom hanks but also mister beast when he
00:11:32
studies might be less no one i don't know but tom hanks i went was immense
00:11:37
uh it is gonna that there was a defect cover of
00:11:40
in a uh promoting a a dental plan even a there are
00:11:44
a agree to that and uh and also uh to be something on its that run to set that you didn't consent to these
00:11:50
but is to show how easy it is to work to generate uh this kind of contents
00:11:56
uh so in that context is not to one foot i would say of course
00:11:59
we have a for the where it's uh so you only fleetingly and the followers
00:12:04
but it could be much more harmful when you use the
00:12:07
fake in the context of elections of for elections purposes for example
00:12:12
and then i found a paper on i don't know uh uh if you are way out this paper uh i'm very intrigued by
00:12:18
it's it's about a man something on says uh it's it's it's
00:12:23
it's a survey that's that's what you twenty thousand papers and patents
00:12:27
that where in computer vision r. e. r. and they say that's the problem is
00:12:32
the way they describe also human they describe humans in these studies ask you object
00:12:38
uh so this article maybe not
00:12:42
a mm completely um uh how
00:12:45
to say it it might be against or um their their research
00:12:50
but you can read the research and i think it's very important as we such
00:12:53
a and that's why we took as well uh technology neutrality that's where we have that
00:12:58
that's everything that you communicate will have an impact on the way your research would be perceived
00:13:04
and uh it's the same when you select the data but in that case it was to show
00:13:09
that most of the of the research that was done to so maybe human beings will to develop
00:13:15
projects a fuss audience but this is so that that's also very important to to and then it's
00:13:20
and then the the the code for more interdisciplinary the a
00:13:24
teens uh when you do a project also to be aware
00:13:27
of the impacts from manic the contest because you only got perspective of research uh that that shouldn't
00:13:33
but i wanted to give you a bit more a background on what
00:13:37
by image with that are are as well that from a legal perspective
00:13:41
because all these technologies the process that's not data on it it depends
00:13:46
uh you'd be can be uh across the ice that's not that's out
00:13:50
so they are physical physiological of the of your data that can be linked to unidentifiable in d. v. d.
00:13:57
or so that's that's how we define with a specific function i can come back to that uh later but there
00:14:04
why it's also so problematic about this but i need to get on biometric technologies
00:14:09
is because they capture information about our identity not only
00:14:13
civic identity but we are how we behave this way
00:14:17
so it's it's about something that can be very person or something
00:14:21
uh which which to which might not want to disclose as well
00:14:26
uh and that's why we have all this discussion about uh we don't we should rely more on what the
00:14:31
biometric that essay rather than what people say and speech
00:14:35
in the cortex of a a new direction for example the
00:14:38
clam identity yes use the f. format ties the identity
00:14:43
uh and we know that the only people is ages of
00:14:46
the technology so uh asking with their uh it's this top yeah we took yeah it doesn't make sense because that
00:14:52
or but they we already use these technologies but it's more how should we use
00:14:56
this technology responsibly and which kind of you shouldn't be allowed as well of these technologies
00:15:03
uh for us it was reinforcements but it's more and more commercial purposes as well and that's why we have
00:15:09
to keep that in mind because the purposes i'm at the same commercial purposes so so with the profit in mind
00:15:15
it's not a law enforcement uh it can be of course to secure the big interest
00:15:21
but it's completely different but this is uh that that that that i thought are completely different
00:15:26
and then we have some benefits that have to be put in
00:15:28
that ends this coming on security and i was thinking as well
00:15:32
as in a case of body identification you could think that sometime
00:15:37
if someone is register some way ah as is our fingerprints register somewhere
00:15:42
it could be very useful to use this information to identify someone who doesn't have an identity
00:15:48
so you mean user so janey i know with that but but we can at a and fingerprints identification we not
00:15:54
be the only technique but it's one of them and we cannot say because we see that the awesome fundamental right issues
00:16:01
that we should not use it in some cases as well i think the in cases of body identification this is very important
00:16:08
as unless lessons are just uh somewhere in in the database can be passports or that type of the that
00:16:13
the bases and of course the reef and the fact that you will lose control over u. one b. d.
00:16:20
you can be profile you can i begin decent inferences incorrect inferences
00:16:24
as well or something that you don't want to be in far from
00:16:27
from your body from your behaviour and you are not necessarily the
00:16:30
way out or you will not have to cooperates a with it so
00:16:35
i'm i'm talking about is to give you some background on the issues because
00:16:39
i think they and there is a lot actually uh to say about this topic
00:16:42
but there's still want to be that i want to raise technology neutrality and
00:16:47
the reason is because for many of perspective when we talk about technology neutrality
00:16:53
tries it's very easy we say we drive generalisation like
00:16:57
the g. d. p. r. with no technologies in mind
00:17:01
yes you switch right uh the stations basically technologies because the answer uh a specific technical problem
00:17:08
but this is not what i wanted to raise here here it's about without
00:17:12
technologies are neutral on that and when you design then when you develop them
00:17:17
and what is interesting is the position of the commission so i'm i'm showing the is to
00:17:23
in the white paper are but also in their uh your pen strategy for
00:17:27
that are so these these two papers off from twenty twenty the commission consider that
00:17:33
uh these so this regulation there should be a technology called
00:17:38
a new toward because we should not attach any value to technologies
00:17:43
because technology is not good it's not bad it's just the news at ten
00:17:48
and there are some without the captain enjoying that and they agree with them it was as seventeen
00:17:53
that and i i wanted to mention one of them is inclined i'm gonna look kind point now you
00:18:00
write something very nice it was uh almost twenty years ago about fisher recognition for me that was
00:18:08
an eye opener because he said that actually the way at that time and see this technology web design
00:18:14
you will exclude some people people will not be able to be recognised because you didn't choose them in
00:18:20
the in the training and the um the training but that's out there was no that's out about specific yet
00:18:26
group of people for example uh um and it's it's very
00:18:30
important to disclose the technical choices that you make but they can
00:18:34
use a political choices so sit agencies to inform and it's
00:18:38
part of fairness as well that it it's part of actually see
00:18:42
uh and it will avoid bases as well so look kind for now uh for me is is one of the authors that's uh
00:18:49
that's what's the um when with something visionary about what is happening uh and now c. could
00:18:55
it does create x. so been in your design i you will make some choices and explain them
00:19:01
uh you find all the authors and that i have more political stance uh
00:19:05
for example p. c. which if you want to read you wrote about the matrix
00:19:10
with these technologies and by all politics because he explained that this technology we have developed as well
00:19:17
um we just basic purpose mine nineties by image speaker identification technologies
00:19:22
it was in in in the context of dominating some populations and
00:19:28
then because of that but they will not trained on the right as
00:19:32
well that ah so is e. e. talk about whiteness of biometric template
00:19:36
at that time so it was in two thousand and that's about that
00:19:39
and if that was an interesting perspective it doesn't say that i agree with you that it's just about
00:19:44
putting some context and perspective and that that's a researcher as a design of ages
00:19:49
design on this side and they don't don't think how their technologies can be used
00:19:54
uh so that's a another point and of course the issues of
00:19:58
biases are so the easiest one is the technical uh races you say
00:20:01
oh you just introduce diversity in your training data set and you are and us garden but that's not the case we know with x.
00:20:08
these bases they also ingrained in our society into when we
00:20:12
can see as well uh uh maybe some people or or
00:20:16
we treat a disability i don't know you you can think of what actually tore at historic
00:20:22
ave is is that we have and you can find them in the systems doesn't mean that
00:20:27
the designers of the researcher i'm not biased but they were present society as
00:20:31
well and this type of bay is easy you don't necessarily aware of them
00:20:36
so there is an important work to be done on that uh about uh making
00:20:41
a visible the bases that you can have any this is very difficult activity has them
00:20:47
and then uh and last but not least a division of really actually see the phillies if he doesn't uh
00:20:55
and uh and you also have that disease which is interesting it's about the
00:21:01
value double chart so what explain he say that technology is they can be
00:21:07
when they are develop either good either but it's it's like
00:21:11
any click on and then depending on which type of technology developed
00:21:16
you will have to do the good side either hire or the the back side
00:21:20
for example you become part of a cookie machine and you say well equipped machine
00:21:24
it's not many a design for good the bases you will not use a coffee
00:21:28
machine to our someone of course can be using that object but that's not but
00:21:33
if you think of this word how can you say that this was not be
00:21:37
designed to on someone uh to keep someone so i like this idea of these uh
00:21:42
value of more value of the bar chart because it shows that
00:21:47
whatever you do it's always good to think about the uh the
00:21:52
are we could be misused it doesn't mean that it imposes more variation of
00:21:57
course on researchers on because you are not here to solve all the problems
00:22:01
but uh it's importance and then to question and to be critical but
00:22:05
with which uh what you do and then uh i just wanted to
00:22:12
so uh i'm not sure you can relate it it's it's but friday what
00:22:15
you watch newspaper that's the technology is either at rest in the neutral stance
00:22:21
not least because every technology is always design according to some implicit and explicit values
00:22:27
by some people or for some people we didn't get to and we we could still in mind
00:22:34
for some uses right um then with us with a four then season constraints and so forth
00:22:38
and i think this is a good summary for me at at the semis uh the issue
00:22:44
and then it's it's an open question i don't have the answer bits about what is your
00:22:49
whatever research on the wall of designers as well that we should not only focused on the use
00:22:54
saying the problem is only the use is that some technology will always be design
00:23:00
to our right out into head out to perfect a society uh in general so after
00:23:06
this uh um um i i would like to to jump it between two to a
00:23:11
division into don't act because i think that's what you would like maybe to to here
00:23:15
i just a few words about it and to tell you while it's a difficult to reach
00:23:19
an agreement on it so uh we have currently the g. d. p. r. on for a in
00:23:26
um the regulation of biometric data with the obvious than data
00:23:30
but the g. d. p. r. there's much regulate with technology themself
00:23:34
and there was a gap that was no regulation for the technologies for
00:23:38
the development design of the use of these uh uh a i says them
00:23:42
and and the commission phone but beware to um to difference research uh and
00:23:50
you shouldn't let you call the way they did it but the first talk use
00:23:53
on more um we ski a risk issue likely ability shoes it was that productivity they
00:23:59
had that in mind when uh when the approbation uh uh the uh the city proposal
00:24:04
uh before the be sure digital proposal sorry when when when they made their web attractive it
00:24:09
and then you have a lot of the provision that are more uh security fifty oriented and you have another part and
00:24:16
it's only about biometric technology displays interesting it's about the impact
00:24:20
on fundamental rights uh and the men folk use they mentioned technologies
00:24:25
so the uh n. b. proposal was publishing twenty twenty one
00:24:30
and in case you don't know the commission just explain to you with this
00:24:34
risk based approach for categories of the ice them so you had nice unacceptable risk
00:24:40
under which for a new remote barometric institution system in two weeks place for law enforcement in real
00:24:46
time but i will come back to then you have high risk much most of the systems here finally
00:24:53
it's very long but i have a slight to show you the and then you need
00:24:57
to raise with the way it was it was even better conditions a bit strange on and
00:25:01
it critical here because for example the commission consider that depicts we were just limited every six
00:25:07
then subject to transparency that you should one use a bad guy interacting with the to pick
00:25:15
but we know for me defective right with the destruction of technology of techniques that not develop
00:25:21
um with an arm in mind because they were developed for a very
00:25:25
specific purpose uh at the origin doesn't mean that we cannot find out up
00:25:29
of there would have been if it's uh to the techniques technologies but anyway
00:25:34
the the commission didn't understand the and the risk that effect goes to society
00:25:40
it it which angela in the negotiation and neither rates well actually biometric technology
00:25:47
use for verification purposes they don't put in your risk according to the commission
00:25:52
so the on that subject any any roads which means that even have
00:25:57
developments for the moment but remember this is that's a lesson exception of light
00:26:02
do we not subject to specific rules regarding the development or uh when they are sold on
00:26:06
the market because the this yeah acting as far as well uttering a new markets also it's
00:26:13
for a a system that will be design i would say you and on top of the hill market or
00:26:19
designing the you of course so yes i wanted to
00:26:23
show you these three types of biometric system that's uh that
00:26:27
are regulated emotion recognition a dime to get the graduation
00:26:33
says that's so according to the commission on these two
00:26:36
sister and they are also limited risk except as some
00:26:39
of the few few cases in uh in particular law enforcement
00:26:44
and remotes biometric identification system so this last one is actually
00:26:50
a broad category that covers feature recognition technology for those gate technology
00:26:57
any type of technology that you could deploy yes when the public spaces um this definition
00:27:03
is far from being perfect it has been high decrease i because from a technical perspective uh
00:27:09
it yeah distribution is always remote at the t. i. thing what i meant is the
00:27:14
remote capture so there is no corporation and there is no one is it's at a distance
00:27:20
um this will not be the the final definition that but is just to
00:27:25
show how far sometimes illegal perspective we
00:27:29
are describing as well technique or the technologies
00:27:33
uh it it's highly difficult because it's which kind of definitions should you should you d. but beside that's what
00:27:40
i want to say and afterwards i will tell you more about the screen with parametric institutions estimates about the calendar
00:27:47
it's important because uh this a. i. x. is not
00:27:51
yet legislation it it may be noted it was proposing
00:27:55
up with twenty twenty one by the commission the concealed
00:27:58
better present good romance either disposition in november twenty twenty two
00:28:03
but your current balance better presents or the citizens of the
00:28:07
e. u. i doubted his position in june twenty twenty three
00:28:11
based on that position a lot of media reported that the a. i. act
00:28:15
was that it this is not true so currently we don't even know if
00:28:19
you would have an a i act we have um the intent is to
00:28:23
additional negotiation that record created so we the three um a institution that's started
00:28:31
and for the moment we haven't been able to agree on a remote magnetic
00:28:36
institutions is there a motion recognition i categorisation system the parliament at as ask
00:28:43
us way to need the use of initial recognition to to have a cheek
00:28:48
purposes because of the respect rallying to
00:28:51
uh uh waiting emotion and operating emotion
00:28:55
um so for the moment we don't know at their uh remote biometric identification stands and i apologise
00:29:02
for that because i have some slides but i really want push to explain explain to the differences
00:29:08
so the commission propose a band so when we saw that the permission
00:29:12
to put a band for the use in real time so like shoes
00:29:17
of this technology in public spaces by put his uh ever thought oh it's it's a good idea
00:29:23
but then you have three broad exceptions and when i separate
00:29:27
exception is because the first one is about finding button show
00:29:31
the victims of crime including missing should run so you would
00:29:35
deploy this technology in real time which which seems to be um
00:29:40
also a disproportionate prevention of union for to live and safety director effect
00:29:47
back so i and he's not national security as well as i think uh
00:29:51
maybe they are using this technology for this purpose is that we don't know it
00:29:57
but the worse is the last one it about depicting localise in prosecuting kind of fun there i suspect
00:30:03
based on one of your fancy that isn't the least of very to
00:30:06
a fancy is uh that can be subject to the european r. s. reference
00:30:10
but this list of references have been interpreted differently
00:30:14
by member states which mean that might know a advances
00:30:18
couldn't it to uh uh your pen nice weren't done on the basis of that you could use this technology
00:30:25
okay so i'm trying to work too good bit faster are um when you're
00:30:30
gonna use of this technology will be high risk and this includes a if a
00:30:34
private company is doing actually a security on behalf of law enforcement in the same
00:30:39
condition the wheel it wouldn't be banned according to the commission would be at risk
00:30:45
so the court seal which are present government when
00:30:48
fat going exception and it's not surprising because the conceit
00:30:52
wants to uh of the components they want us sure
00:30:55
more civilians so it's using this technology for most audience
00:30:59
the high for example or prevention of attack to critical
00:31:03
infrastructures so it shows that as soon as you start opening
00:31:07
these a box of exceptions you can go very far
00:31:11
um and then the buttons and and the parliament it's the
00:31:15
citizens the parliament does not want any use of this
00:31:18
technology in real time or a past event or sort of
00:31:23
white prospective use by any actors they still one exception
00:31:28
is in case of the proxy cushion of specific serious crimes
00:31:32
after judicial but isolation this is the only exception um
00:31:37
but the parliament's heart difficulties to reach its own a political
00:31:43
consensus on these documents only supposition just before beef and
00:31:49
be either t. deposition uh the political party say that they
00:31:53
wanted more exception on the screen with a metric units patients them and the problem here that's why i say easy
00:32:00
to please the guy because we have to find a balance
00:32:03
between between the predicted the the substantial public security uh interest
00:32:09
and so it's it's the security i mean did yours and their rights are fundamental rights in this book as bases
00:32:17
and another problem that i see is that the commission as the justify this exception because
00:32:23
they are impacting our fundamental rights and there is no we've either is that these uh technologies
00:32:29
our efficiency that they are necessary in our uh democratic critics bases there is no evidence
00:32:36
uh and you have several and use that are against any form of user these technologies
00:32:42
but since i'm a member state are already using a mm uh by magic with
00:32:47
the are really using facial recognition technologies it prospectively for investigation purposes i don't think
00:32:53
that the conceal watson the government but they want to ban it we can i
00:32:59
think they can and then it so now we are really at um at the crossroads
00:33:05
of course i mean what would happen to these um a. i. x. because as it stands
00:33:10
now we don't have an agreement on that the european parliament we have election next year in engine
00:33:16
and if we don't have an agreement by the end of the year it's very difficult
00:33:20
to ensure that we we don't agreements uh uh the first months of the twenty twenty four
00:33:25
so it's it's it's a very difficult the balance to find any to put a
00:33:30
good balance and yet so it's not going to be fundamental right buses predicts security
00:33:35
is going to be a a different provision of the act that
00:33:39
will be a compromise uh oh that will be balance against each other
00:33:44
and then to finish i just wanted to give you a one update
00:33:47
it's about uh your print gotta human rights concerning the the right to privacy
00:33:53
very interesting case and that was published in july where a new but
00:33:58
administrator was our estate because he was basically a demonstrating in the underground
00:34:04
and after a few a few days i think the police on this trace again in the in the underground
00:34:11
the use facial recognition retrospectively in life because they could identify him
00:34:16
through a social made yeah uh she's features the courts it's not necessarily
00:34:21
a democratic society at the this kind of use that they say
00:34:24
that's for that context been may be they didn't take position with their
00:34:30
deploying fisher recognition technology in public space he's
00:34:33
he's up isn't that's a necessary in democratic society
00:34:38
that's the way that they they believe to the e. u. institutions that that's a political debate
00:34:43
and that's a debate as well that we should all have as citizens
00:34:46
that we are not attributes because we did agape that's to the institution
00:34:51
uh then i know that in sweden and you you might have the
00:34:54
right to um oh yeah i found them you can me participate more in
00:35:00
this type of the bait but it's also to think of which kind of
00:35:04
society would like to even as a citizen that as as researchers or uh
Conference Program
(Keynote Talk) Privacy-Preserving Machine Learning : theoretical and practical considerations
Prof. Slava Voloshynovskiy, University of Geneva, professor with the Department of Computer Science and head of the Stochastic Information Processing group
Oct. 11, 2023 · 7:55 a.m.
2514 views
5 minutes Q&A - Privacy-Preserving Machine Learning : theoretical and practical considerations
Prof. Slava Voloshynovskiy, University of Geneva, professor with the Department of Computer Science and head of the Stochastic Information Processing group
Oct. 11, 2023 · 8:40 a.m.
Enabling Digital Sovereignity With ML
Vladimir Vujovic, Senior Digital Innovation Manager, SICPA
Oct. 11, 2023 · 8:44 a.m.
5 minutes Q&A - Enabling Digital Sovereignity With ML
Vladimir Vujovic, Senior Digital Innovation Manager, SICPA
Oct. 11, 2023 · 8:58 a.m.
Privacy-Enhanced Computation in the Age of AI
Dr. Dimitar Jechev, Co-founder and CTO of Inpher
Oct. 11, 2023 · 9:01 a.m.
139 views
5 minutes Q&A - Privacy-Enhanced Computation in the Age of AI
Dr. Dimitar Jechev, Co-founder and CTO of Inpher
Oct. 11, 2023 · 9:20 a.m.
Privacy by Design Age Verification & Online Child Safety
Dr. Onur Yürüten, Head of Age Assurance Solutions and Senior ML Engineer in Privately
Oct. 11, 2023 · 9:26 a.m.
5 minutes Q&A - Privacy by Design Age Verification & Online Child Safety
Dr. Onur Yürüten, Head of Age Assurance Solutions and Senior ML Engineer in Privately
Oct. 11, 2023 · 9:41 a.m.
(Keynote Talk) Biometrics in the era of AI: From utopia to dystopia?
Dr. Catherine Jasserand, KU Leuven (Belgium), Marie Skłodowska-Curie fellow at Biometric Law Lab
Oct. 11, 2023 · 11:06 a.m.
5 minutes Q&A - Biometrics in the era of AI: From utopia to dystopia?
Dr. Catherine Jasserand, KU Leuven (Belgium), Marie Skłodowska-Curie fellow at Biometric Law Lab
Oct. 11, 2023 · 11:42 a.m.
AI and Privacy
Alexandre Jotterand, CIPP/E, CIPM, attorney-at-law, partner at id est avocats
Oct. 11, 2023 · 11:48 a.m.
5 minutes Q&A - AI and Privacy
Alexandre Jotterand, CIPP/E, CIPM, attorney-at-law, partner at id est avocats
Oct. 11, 2023 · 12:06 p.m.
Preliminary Pperspectives on the Ethical Implications of GenAI
Julien Pache, A Partner at Ethix and Venture Partner at Verve Ventures
Oct. 11, 2023 · 12:12 p.m.
5 minutes Q&A - Preliminary Pperspectives on the Ethical Implications of GenAI
Julien Pache, A Partner at Ethix and Venture Partner at Verve Ventures
Oct. 11, 2023 · 12:30 p.m.
AI & Media: Can You Still Trust Information
Mounir Krichane, Director of the EPFL Media Center
Oct. 11, 2023 · 12:32 p.m.
5 minutes Q&A - AI & Media: Can You Still Trust Information
Mounir Krichane, Director of the EPFL Media Center
Oct. 11, 2023 · 12:54 p.m.
(Keynote Talk) Unlocking the Power of Artificial Intelligence for Precision Medicine with Privacy-Enhancing Technologies
Prof. Jean Louis Raisaro, CHUV-UNIL, assistant professor of Biomedical Informatics and Data Science at the Faculty of Biology and Medicine and the head of the Clinical Data Science Group at the Biomedical Data Science Center
Oct. 11, 2023 · 1:22 p.m.
5 minutes Q&A - Unlocking the Power of Artificial Intelligence for Precision Medicine with Privacy-Enhancing Technologies
Prof. Jean Louis Raisaro, CHUV-UNIL, assistant professor of Biomedical Informatics and Data Science at the Faculty of Biology and Medicine and the head of the Clinical Data Science Group at the Biomedical Data Science Center
Oct. 11, 2023 · 1:50 p.m.
Genomics, AI and Privacy
Julien Duc, Co-Founder and Co-CEO of Nexco Analytics
Oct. 11, 2023 · 2:01 p.m.
5 minutes Q&A - Genomics, AI and Privacy
Julien Duc, Co-Founder and Co-CEO of Nexco Analytics
Oct. 11, 2023 · 2:18 p.m.
How trust & transparency lead the success of an Idiap student's Master's project in fraud detection
Raphaël Lüthi, Machine Learning and AI Lead at Groupe Mutuel
Oct. 11, 2023 · 2:22 p.m.
5 minutes Q&A - How trust & transparency lead the success of an Idiap student's Master's project in fraud detection
Raphaël Lüthi, Machine Learning and AI Lead at Groupe Mutuel
Oct. 11, 2023 · 2:38 p.m.
Recommended talks
Management of health data in an industrial company
V. Srikanth Nallanthighal
Sept. 5, 2019 · 1:57 p.m.
146 views