Player is loading...
5 minutes Q&A - AI and Privacy
Alexandre Jotterand, CIPP/E, CIPM, attorney-at-law, partner at id est avocats
Wednesday, Oct. 11, 2023 · 12:06 p.m. · 05m 24s
Embed
Transcriptions
Note: this content has been automatically generated.
00:00:00
and i'm yes
00:00:01
questions
00:00:01
um
00:00:07
hello so
00:00:08
thank you for the talk a um
00:00:10
i think the question are you in your legal bubble discussing about
00:00:16
uh whether the machine learning model is considered personal these or not
00:00:22
what are the the mentioned adding modal yes because in some way i'm
00:00:26
gonna show it later in the talk but it's it's a sort of
00:00:30
compressed way of your training the right so it might be also considered this personal data
00:00:38
yeah um is when we start sharing models instead of the uh
00:00:43
the question is uh are we under g. d. p. or not
00:00:46
yeah and the the easy answer to that is that you know whenever
00:00:51
you could indirectly identify individuals within your that assets which
00:00:55
as soon will be possible in a lot of cases
00:00:58
then you'll match unloading model will contain person of that that
00:01:04
and that's the answer you will get from a new lawyer and the and the but more importantly uh courts
00:01:09
basically that's why they would say they will not really focus on the tech aspect that goes beyond that
00:01:16
um um
00:01:19
um um um
00:01:24
like your problem yeah my
00:01:26
side just related to that um is the definition or personally
00:01:31
identifiable information a static definition or does not involve
00:01:35
with what is technically capable or possible at any time
00:01:38
for example maybe nowadays you could almost identify someone by the slayer which directs for example right
00:01:44
maybe this was all possible before so how do we see that the definition of these uh
00:01:50
things evolve with with the capabilities or with ah yeah that's yeah okay
00:01:55
so it's definitely not static at and it first of all it depend on the context all
00:02:02
what is personalised that for you may not be a person and that that for me
00:02:06
so that you know directed them fire there would be person that that for every on
00:02:11
and then they are in directing the fire maybe you have the additional
00:02:14
information that is required to you identify a other individuals and i do not
00:02:20
for instance you have you you on site for your data i use and you so
00:02:25
suddenly mice that that's that's for you would still personal data because you have to actually
00:02:31
to uh and to link back the individuals uh the that that to the individuals but i don't have it
00:02:38
so for me it's not personally does for you it's not for me which have a lot
00:02:41
of impact and you know all we must both be hate on the that the protection those
00:02:47
and the question for me is whether i could still you know you don't if i
00:02:51
the individuals with but that our the information that i have in my surroundings
00:02:56
and with that you have to take into account the reasonable
00:02:59
means that i could help page to identify the individuals and and
00:03:07
i don't i don't think there is a lot of research from the legal fees
00:03:10
it now that goes into you know how far do you need to look at
00:03:15
so we'll more consider uh basically the current coupon and why we will say that
00:03:20
and maybe you know if you really have to be strict about
00:03:25
proving that what you have is really not person all the time
00:03:29
does answer your question sure yeah i'm
00:03:38
okay so a question is what is from a legal perspective difference between let's
00:03:43
say google and cared deeply t. both all systems take next do you some output
00:03:49
they have it asked is relevant what's what is the difference from the legal
00:03:53
stand yeah i think it's a very it's a very sound questions uh i am
00:03:58
and um so i guess the view off the get and paint the case i mentioned was
00:04:03
that while it is the same what this i don't want to re open the door you know
00:04:08
uh creating a situation where a a search engine would not be lawful under a new
00:04:12
law because they're doing kind of the same thing but id we you doing the same thing
00:04:18
and i'm not sure uh and i i'm not entirely sure with that as well because you know the impact
00:04:23
but tragically taken have an individualist is very different
00:04:26
than the one and we will search engine has
00:04:36
i i yeah well one of the differences that's with school so it the ruling
00:04:45
of the court of justice actually came also and with the right to be forgotten
00:04:51
so they say okay you you are legitimacy in collecting the that task right
00:04:55
uh scrapping does that than line and and creating these you know inventor yeah whatever
00:05:01
and but in exchange you will have quite cumbersome obligations which will
00:05:05
be to the to make the press and disappear from the results
00:05:10
whichever challenging at the beginning and as far as i'm concerned
00:05:14
i'm not sure that judge if it is currently able to do that
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.
2509 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.
138 views
