Maximise Player Approvals with AI-Powered Fraud Fools

0:02
Good afternoon and welcome to the latest SBC webinar, maximizing player approvals
0:08
with AI powered fraud decisioning. We’re looking at the balance when it comes to
0:12
onboarding responsible gaming and how we can integrate and implement AI into this
0:17
process in the modern gambling ecosystem. Before we get going, I would
0:22
invite you to ask any questions that you have on the subject or any follow-ups to
0:27
what our panel say because we’ve got a great panel. So yeah, feel free to pick
0:31
their brains. And on that, I will let the panel introduce themselves.
0:35
Stephanie, if you want to start just by telling us who you are and you’ve been
0:39
on the SBC webinar before, but remind the audience who you are and yeah, where
0:44
you’re from and what you do.
0:46
>> Yeah. Um, always a pleasure. So, thank you for the opportunity. Uh, my name is
0:50
Stephanie. I am a product um PMM at SIFT, a AI fraud decisioning solution.
0:56
Um, I sit between uh product and our customer success. So, it’s really my job
1:02
to get into the nitty-gritty of uh the use cases here in I gaming and make
1:07
sureing making sure that the innovation stays in line with helping those um um
1:12
solve for that. So, one of my favorite industries and the people in it always
1:16
fascinating.
1:18
>> Brilliant. And the nitty-gritty is what we shall get into today. Uh Peter uh
1:23
yeah next up
1:25
>> of course so thanks for very much for inviting me for the webinar and uh
1:30
successful cooperation with SBC and uh it’s not my first event but uh very glad
1:36
to to participate. So by way of introduction uh I’m the MLRO for um the
1:43
Netherlands market and French market at FDJ United. FDJ is is is a large
1:49
European operator operating online and lottery businesses in in in in Europe.
1:55
So uh I am based in Malta uh with my experience you know uh spanning
2:01
different industries. So gambling uh also financial services before. So um
2:07
let’s say I am also apart from let’s say my professional activity I am heavily
2:12
involved in uh some industry industry uh associations industry bodies such as a
2:20
camps which is basically um the the the the uh professional professional uh
2:27
association for compliance officers focused on AML task. So, as the local
2:32
board member there and uh yeah, looking forward to to to exchange ideas during
2:37
the the the webinar.
2:39
>> Brilliant. And uh Greg, uh last but certainly not least, uh yeah, let them
2:43
know who you are.
2:45
>> Thank you, Joe. Uh really nice to be here. Thank you very much. And sharing
2:49
some really nice uh views on on the topic. So, quick intro intro from my
2:55
side. the chief risk and compliance officer of play north which our main
2:59
market is Netherlands with the brand Canino and before Play North I was five
3:04
years with Bedway and then before Bedway I was almost nine years with Kaizen
3:09
Gaming. So in multiple operational risk and compliance roles through the last 16
3:15
years and you know good to be here guys
3:20
>> most certainly and let’s get into it. Um, and once again, I remind you to, if
3:23
you’ve got any questions, please feel free to fire them in. But let’s get into
3:26
it. Um, and I feel like when we have these discussions, it’s kind of
3:31
challenging because we we’re talking about a general subject, but it varies
3:35
market by market. A new approach is needed for each different market and
3:39
largely that is down to the regulatory frame, the regulatory framework and
3:44
ensuring that you are being compliant when it comes to on boarding. Um, I
3:49
wanted to kind of pick your brains and just get get you to start off just by
3:53
telling me uh any markets that you think have gotten gotten it particularly right
3:59
when it comes to the on boarding framework. Uh, Greg, you you were last
4:03
with the intro, so if you want to be first up here.
4:06
>> Yeah. So I mean I can speak for for Netherlands which is
4:11
quite an interesting onboarding process where the customer needs to verify their
4:17
bank account, their ID and their documents go through certain checks with
4:23
the national self-exclusion database which is called crooks and the CCBR
4:27
which is the database for um bankruptcy and guardianship and then and
4:34
then obviously set set certain limits before they are able to play and then
4:38
obviously get one more step with the contact point. If these limits are one
4:42
more step is is is the process that we need to to follow. There is, you know,
4:48
continuously need to be optimized every step of the way. following it really
4:53
really closely with reports and alerts and keeping a very close eye on where
5:00
there is dropouts at what point which tools are performing well and which not
5:06
and then obviously acting as fast as possible with really tight and close
5:11
SLAs’s utilizing
5:14
expanded technology like AI at some point in order to be pro proactive and
5:22
take care of UX uh uh customer UX plus maintain compliance at all times.
5:31
>> Interesting. And I guess it’s really important that the process uh and the
5:35
tools don’t stand still. They continue to evolve off the back of the data that
5:39
comes from the RG tools.
5:42
>> Definitely. Definitely. And it’s and it’s a it’s it’s an ongoing process not
5:46
just from myself but uh from a team of more than 15 people that uh daily
5:51
analyze the trends pay attention to the smallest details and try to optimize
5:58
internal and external tools through uh contact with the vendors and obviously
6:04
in that process I don’t even include the anti-fraud tools that we have uh and
6:10
that any every company needs to have nowadays. is in terms of uh protection
6:15
and having a really clean database in terms of uh various sources of uh you
6:21
know malicious activity.
6:23
>> Yeah, that’s uh fascinating and uh we are going to explore that further but
6:27
before we go into it uh Peter any uh any particular markets that you think have
6:32
gotten it right or have even gotten it wrong in their framework? Well, so so I
6:37
can concur with uh Greg here that Netherlands with these kind of
6:42
sophisticated on boarding tools which involve u of course contra account um uh
6:50
creation uh electronic verifications, crooks databases plus of course in many
6:56
occasions uh you could use for example the you know very strong uh indicator of
7:02
authentification uh thing as the citizen citizen number of of of of a person.
7:10
This creates a very stable and let’s say um environment which also influences um
7:17
let’s say uh minimizing uh let’s say the damage or fraudulent registrations so to
7:24
speak. You know if you look uh even at the statistics let’s say like uh some
7:29
operators um historically who would operate in the Netherlands let’s say uh
7:34
within what is called grayson zone you know under other licenses like Malta and
7:38
so on before the local regulation uh if you look at the statistics let’s say uh
7:45
the there’s they’re words apart in terms of you know um uh
7:51
fraudulent registrations and so on. So this this framework is really effective
7:56
and even uh before let’s say fully applying the ADAS you know the the the
8:01
European electronic ids and so on. So so so so I must say that um also this is
8:08
the ecosystem which has been in the Netherlands uh for uh some time already
8:13
you know between different banks and uh also high adoption rate of tools which
8:18
uh which are connected to things like open banking uh within Netherlands uh
8:23
and also the culture which which which is in that jurisdiction which enables
8:27
the use of such tools is I mean it’s really a good good uh plan for
8:33
successful on boarding of of the customers. Um the second part of the
8:38
question just to answer Joe here uh about if there are some markets who got
8:43
it wrong. Maybe I would not be so strong in the words you know like condemning
8:47
the countries that it’s wrong. It’s basically difference of approaches and
8:50
difference of culture. You may have you know societies where adoption of
8:55
electronic means um of you know uh familiarity with the tools is higher due
9:01
to you know the cultural uh and also some political decisions as well but you
9:07
have countries which are more traditional
9:10
let’s mention for instance France where you have more traditional way of um you
9:15
know creation of the account you know um you don’t really have um let’s say
9:21
necessarily All these electronic verifications and uh registration tools
9:25
and uh the whole account let’s say verification process may take you know
9:31
not an instant but uh couple of weeks in fact uh which may be the case. So this
9:38
is very heavily influenced by by by the culture of the of of of the technical
9:43
adoption uh in the country and also the banking uh ecosystem because let’s
9:48
remember that in on many occasions such electronic registrations are facilitated
9:53
by the use of uh regulated financial institutions or other regulated
9:58
businesses you know which already verified uh the identity and of course
10:03
you know uh there are some more traditional countries uh I can speak
10:07
also So from experience you know in banking sector you have countries like
10:11
in Luxmbborg I mean Luxmbborg does not have its own license let’s say but um
10:15
but uh in the sense of you know adoption of of modern tools uh by the banking
10:21
sector maybe a little bit you know um uh not as advanced as as as let’s say
10:27
Netherlands. Okay. Yeah. That’s interesting. And uh
10:32
it really is important um that you that the cultural nuances of the market are
10:36
understood and also a collaboration with the financial sector is uh is really
10:40
crucial tr too and Steph to to round us off on that one. Um yeah talk to us
10:46
about a market that has been of particular interest to yourself. I you
10:49
know I I think the Netherlands and the has been kind of the gold star, gold
10:54
standard, but um what I’d like to double down on too is when I look at operators,
10:59
your job is not easy because as you deal with different markets
11:03
um the the digitization of the ecosystem, right, is really important in
11:10
where the regulators define compliance. And what what I find difficult for
11:15
operators is that um they are expected to provide the speed, you know, uh
11:23
verification perhaps like the Netherlands, but they’re operating in a
11:26
situation where they don’t have that foundational um um access to that type
11:32
of information. And so what I find in um when I look at the different markets and
11:37
the regulations is that some regulators look at gold standards and try to imply
11:42
kind of the same requirements. Um and from a player perspective, you know,
11:47
people have less and less patience and so they want that instant kind of move
11:53
forward. And I think um operators have to find that balance. And
11:58
I think you know where is the where where’s the juice versus the squeeze in
12:03
terms of how you deal with regulations is really taking into the cultural
12:07
account of you know the ecosystem where the regulators have in terms of where
12:12
their perspectives are um and the tolerance of the players itself for that
12:17
type of friction. Um so I do think Netherlands is something that has become
12:22
um more and more uh looked at as as an ideal but it is to your guys’s point um
12:30
there’s a foundation already to provide access to that kind of check in terms of
12:34
speed. Yeah, Stephan, if I I can interject
12:38
there quickly is nowadays with all these examples and we also have Portugal which
12:45
the on boarding is super fast and super convenient
12:50
also a lot of countries is that when a regulator starts planning or starts
12:55
reviewing there are already some best case scenarios. So we don’t need to
13:00
reinvent the wheel all the time. when we see a a successful model then it’s easy
13:05
to copy. Obviously some regulations are five 10 years old. So there there is a
13:11
lag and a differentiation. Plus the other interesting uh case here is that
13:16
when when a multi-country operator has to operate in countries with different
13:21
standards and different processes, this gets even more complicated in terms of
13:26
operations, systems, costs, etc.
13:30
>> Yeah. Yeah. Yeah, that’s a great point. And uh Steph, I’m going to come back to
13:34
you there because um just to follow up on what Greg said, um a cookie cutter
13:39
approach like a copy paste approach, it doesn’t work markettomarket. And that’s
13:43
even the case when we’re talking about AI, right?
13:46
>> Yeah. Yeah. It it’s so true. Um you know AI can drive a lot of automation in
13:51
terms of the risk assessments but uh because each region is so specialized it
13:58
becomes very important to look at AI as a upleveling tool um but there are uh
14:07
there is still a very big role for the fraud and compliance teams to specialize
14:12
in how to adopt that particular AI tool especially since regulations are uh
14:18
moving really fast and so um the overall the overall insights of the players
14:26
become more and more important because the AI can drive some behavior trends um
14:31
and give kind of a a more of an instant snapshot of what risk looks like. Um
14:38
however I do think about AI as just one part of the tool set because uh
14:44
operators do need to go in and specialize and actually look at risk
14:48
signals outside of the AI to understand how to um supplement. So I’ll give you
14:53
an example. um in Netherlands maybe there the vast majority will be able to
14:58
provide you know um the connections of the verifications but there are segments
15:03
of groups that had an operator come in and said they’re they’re having
15:06
difficulty trying to verify because maybe the bank doesn’t have that how can
15:10
AI and how can player insights across your internal departments and your
15:15
vendors used to supplement that risk assessment of a player when you can’t as
15:22
easily access the verification And I think this is where the role of kind of
15:27
technology and cohesiveness around operators really start to gain more
15:31
importance.
15:33
>> Exactly. just to to to mention and to add on what Stefani mentioned as well.
15:37
is very important as well how you train the AI models and what data sets are
15:42
used because of course you may create I mean the model can create some sort of
15:46
bias if you for instance have very specific market let’s say with different
15:51
transaction patterns different you know fraud type typologies and AML typologies
15:57
for instance and you know without let’s say adjusting the model you would risk a
16:02
lot uh if you try to implement it directly in other market in the new
16:06
market you which may be completely different. And of course this raises
16:11
under other questions which which concern you know data protection you
16:16
know the use of automated processing of data decision making and so on. These
16:21
are very heavily regulated uh matters in certain markets. I mean in the EU we
16:26
have you know AI act and uh this this this kind of you know uh
16:33
let’s say certain requirements certain prohibited activities uh for this so so
16:38
so definitely as you mentioned I mean AI will not
16:42
eliminate humans of of I mean it would create even more specialized roles in
16:47
this just to add my two cents here
16:52
>> it does and so I think it’s very important for um when you look at new
16:57
tools to want to you know u maintain a certain um level of flexibility and
17:03
control uh we’re not even talking about trends but just the speed of compliance
17:07
right within a month the new populationation come can come in and you
17:11
need to be able to have um a balance between what you can control and then
17:16
where AI plays in that role.
17:20
>> Yeah definitely as as what what Joe mentioned before it’s It can be a cookie
17:25
cutter approach like first of all many many internal teams can leverage of AI
17:31
capabilities and of different AI capabilities from customer services from
17:36
frontline from payments uh and different types of AI like you know in terms of
17:42
analysis forecasting um you know explanation so it’s it’s
17:49
really I think it’s um the industry has been a bit late into
17:54
tapping into that cons versus other industries as we we hear you know from
18:00
from tech news etc. but I think there’s a lot of uh ahead of us in terms of that
18:06
and obviously security and safety is the outmost importance here like in
18:13
terms of securing the data.
18:17
>> Yeah. And we often kind of talk about AI kind of removing human roles, but it
18:22
sounds more to me the sentiment I’m getting from from you guys is that we’re
18:26
going to see new roles coming in as a result of AI uh new human roles. I mean,
18:31
not new AI roles. There’s going to be new job roles for specialists in AI.
18:38
you know, I I I um it is kind of a larger operational shift and I think
18:44
sometimes when uh uh operators look at the cost versus benefit of investing
18:51
into a new technology, it can be daunting, right? You know, uh what else
18:55
do I need to do? It’s not just the widget. Um and I think the mindfulness
19:01
around uh the opportunity to use AI and while compliance and fraud has been a
19:10
has been kind of seen as a hurdle um when you talk about the shift into new
19:14
positions, it could be also um the the evolution of existing roles into um u
19:22
from compliance and and uh fraud which has kind of been looked at you know in
19:28
in a positive way, in a growingly importance way as a revenue um perhaps
19:33
uh slower uh driving revenue to be a little bit slower. But I think the new
19:38
roles or the new roles or the evolution of the roles is really about shifting
19:42
the perspective and taking advantage of that to uh evolve roles to be more of a
19:48
ghost driver. And yes, uh the human element of it is I think more of an
19:53
evolution about how these teams can be more strategic in driving the UX
19:59
strategy instead of just you know these are the policies to stop it but how can
20:03
I use utilize these insights um to to solve for multiple use cases but really
20:10
accelerate the UX part of it. So, not, you know, Greg talked about, you know,
20:14
you got to look at the compliance requirements, but then also balance the
20:17
UX. And I think the next evolution of what the rules will look like, um, is is
20:23
the ability to kind of merge the two, um, and lean towards more of a UX while
20:29
while driving that compliance piece. So, yes, there are new roles that are
20:33
needed. Um because I do think it’s a mistake if people go in and if any
20:38
vendor says you know set it and forget it I will I like
20:43
technology where can I invest in the stock because I don’t think that it is
20:47
um um the right approach especially given the shifting change of the I
20:51
gaming landscape and the compliance landscape. Yeah, I I agree. And um we we
20:56
had a question in, but uh before before we go to that, um uh P, I want to kind
21:01
of pull you back on something you said just before about the uh the the
21:05
European Act and um the European AI Act and uh regulation. Um how important is
21:12
it that uh regulation does evolve to a point where we can use AI in a positive
21:18
way and in an effective way? And I guess are the signs good as well like are we
21:22
seeing things that are optimistic on this front?
21:26
>> Well I mean talking about regulation I think a few years ago it already came
21:31
out from FATF so financials action task force where uh that body elaborated uh a
21:38
paper on uh basically use of new technologies including AI and so on.
21:43
some recommendations not only addressed to the industry but also to the
21:46
regulators and of course from what I have seen uh among different
21:50
jurisdictions. you would have um you know some jurisdictions. Uh this is a
21:54
very I mean AI as a topic I mean has been historically a very sensitive issue
22:00
especially in uh regions such as European Union which really you know
22:04
treasure the data privacy informed consents um you know limitations on
22:10
automated processings. Of course um we are referring to GDPR but not only I
22:16
mean um uh about this and uh some regulators followed through as well
22:23
adopting even themselves AI tools for you know filtering the data uh you know
22:30
planning the inspections for for instance
22:34
some regulators still still still don’t don’t follow through and of course Um I
22:42
can see that some jurisdictions um you could justify the use of AI as long as
22:48
of course you know the use of AI does not fall into you know under this EU act
22:53
what is the prohibited category and what is important for organizations I think
22:58
especially the ones you know based on the size you know the economy of scale
23:03
uh is to have um sort of coordinating function AI governance function. I think
23:10
that this is um something which u you know referring to what Stephanie
23:15
mentioned about creation of new roles or evolution of the of the roles that the
23:19
governance aspect you know what we traditionally traditionally associate
23:23
with policies procedures um you know compliance in wider sense will also in
23:29
uh evolve to include the AI governance governance part and this function uh
23:36
would be crucial in you know uh doing the risk assessment It’s not only you
23:40
know like you have like in financial services you have the risk management
23:44
function which is like compulsory also for you know the listed companies now it
23:49
will be spreading more and more um as a specialized function which would uh
23:54
assess the risk risks connected with the use of AI and this is already happening
23:59
in in in in bigger organizations uh you know creation of such functions and
24:04
documentation of uh the use cases of AI if this is allowed uh what is the the
24:10
level of intern of of risk for the company? uh if the processing activity
24:15
of this XY Z is first of all allowed and then to which category it falls what
24:20
kind of justifications and safeguards have to be put in place and this is more
24:24
of a interdep departmental role um such AI governance and a a new stakeholder
24:31
let’s say new kid on the block for for you know compliance professionals to
24:35
interact with and uh and this is really exciting because it’s happening of
24:39
course you have some jurisdictions um where I mean as you mentioned And there
24:43
will be some prohibited cases where you cannot really use AI just not
24:48
necessarily because of the act but also because of the AML regulations. So we
24:53
mention you know compliance AML professionals like like myself like MLRO
24:58
you know who take decisions. I mean technically speaking uh you could uh you
25:04
know put some algorithmic decision making but of course um this is not
25:09
something which uh the regulators would find you know acceptable that uh
25:13
decisions I made purely by AI. It’s a you know enabler facilitator
25:18
facilitating tool but not let’s say you know unchecked uh automated decisions
25:25
this is the you know the red line for most of the regulators I have seen
25:30
>> and I just wanted to add a comment oh go ahead on that just going back to the
25:35
human part of it and the level like you know the perception of AI is scary right
25:39
it’s an amorphous blackbox thing and so I I I do think a lot of education is
25:46
required but also when you look at AI solutions for your own operators you
25:50
need to be able to see why those decisions were made and so again going
25:54
back to maintaining a level of control and insight um I think is something that
26:00
is important and again I don’t think AI should be a set it and forget it. Um
26:04
it’s really important as you evaluate your AI processes. How can you prove if
26:09
you are brought to you know the court that that you do have control and
26:14
insight into you know how these decisions are being made.
26:18
>> Yeah. And based on what F said like it remind me of how many cases as an MLRO
26:25
in the past or managing fraud in in companies where you’re just looking at a
26:30
case and it reminds you of something like you have seen that in the past like
26:36
this exact same pattern but you can’t really remember when exactly and then
26:44
you start holding it back. So not releasing a withdrawal uh reviewing it
26:50
again. Some things that as you said P it can help it can expedite it can scale
26:56
but also has a lot of pitfalls that we need to be really mindful about when it
27:03
comes to serious decision making that and decisions that have to do with
27:09
regulatory compliance with AML compliance with uh big big decisions. I
27:15
mean I’m also reflecting sometimes you know even for personal use you do some
27:19
analysis with paid versions and you set the parameters and then the AI doesn’t
27:24
read the parameters well you correct it and it says like okay sorry I made a
27:29
mistake well this in our professional realm it can be really the case where
27:34
mistakes are super costly from the bottom line to all the way up to the
27:39
reputation of the business. So I think it’s really really important that uh
27:45
it’s a way to facilitate things but always the trained eye and and let’s say
27:51
our experience it really cannot be um exchanged in this cases.
27:58
>> Yeah. And I mean you know we’re talking about the big decisions here right?
28:01
We’re talking about you know high level decisions but I mean this risk factor it
28:05
does extend to decisions that or you know lesser impactful areas and elements
28:12
of business. uh forgive me for a mainstream kind of reference, but uh the
28:17
the recent story with Grock AI from uh I think it’s like the past 24 hours is
28:23
quite a crazy one and an example of yeah why you kind of need I guess Peter to
28:28
what you’re saying you need to go in and keep tabs and keep make sure your AI is
28:32
set and yeah quite a crazy story though but I I thought worth referencing
28:37
um before I move on to the question um that we’ve had in around the other side
28:42
of this debate obviously uh yeah the the industry can use AI but
28:48
you what happens when fraudsters start to leverage AI as their process?
28:52
>> Um my answer is they already are they you know uh Greg had mentioned that you
28:58
know the industry has been um a little bit slower to adopt um you know compared
29:03
to other industries um and even slower in terms of the fraudsters. uh you know
29:08
I I think for a long time uh verifications
29:13
um is where kind of rooted in terms of uh compliance and fraud has been has
29:19
been um trying to find a safe place but fraud technology or anti-fraud
29:25
technology is becoming is accelerating at a pace that is is is phenomenal. We
29:30
do dark web searches. They use Tik Tok. I I’ve seen it guides on Tik Tok on how
29:38
to get past these synthetic IDs. And so what’s happening in the market is that
29:43
it’s already happening. Um and it’s making verifications while still
29:47
incredibly important. um it makes it only as effective going all the way back
29:54
to the digit dig digitization of open banking and interconnectivity of
30:00
the culture itself. And so fraudsters will find holes in markets where it’s
30:06
not tightly wound and tightly digitized between citizenship and all these other
30:12
things. And so how do you combat that? Um you fight fire with fire. you know,
30:17
they’re moving faster um than ever. And so I I do think where you look at the
30:24
vast majority of kind of how analysis is done, like Greg said, it’s very very
30:28
human-based, very very manual still. I still see folks looking at analysis on
30:33
VLOOKUP and Excel sheets, you know, while the frauders using Tik Tok and
30:38
fake IDs. And so, um, I think the question is, you know, are
30:45
they being used and how to combat that with AI? Um, you’re what you’re really
30:50
looking at for AI is the ability to automate insights as quickly as possible
30:56
and to track trends as quickly as possible. That is how I view AI. um
31:02
because that’s how AI is being used right now is that they are they are
31:06
being used to collaborate internally between um the fraudsters and also um to
31:14
bypass your policies as quickly as possible. U Greg you know before this
31:20
kind of talked about you know the aging of accounts I believe where they will
31:23
just switch tactics as quickly as possible so you can’t set it and forget
31:27
it. Now you cannot have static controls
31:31
basically with of this with this kind of um uh fraudsters because you know uh
31:36
they are always let’s say steps ahead um sometimes light years ahead of you know
31:41
certain businesses including our businesses I mean you remember fraudster
31:45
is a criminal so most likely and you know they don’t have any compliance or
31:50
you know GDPR or data protection you know privacy privacy limitations so and
31:56
there are tools I don’t want to name them of course you But there are tools
31:59
which are very easily accessible and they don’t have any limits you know
32:03
based on the language models you know like chat GDP but without any
32:08
limitations where it sort of influences what is called democratization or you
32:13
know like wide wide adoption by you know a lay person uh to be to be a fraudster
32:19
you know at the scale never seen before. So uh you could uh see as you mentioned
32:24
synthetic ids of course it’s kind of a race arms race between between the
32:29
operators and between uh the frosters you may have you know people I mean like
32:35
what can be the measure to fight I don’t know synthetic IDs okay you have some
32:39
document recognition software but you know also on the other side you know
32:42
synthetic ids they really improve and uh you can uh fake some features of these
32:48
uh like the real documents which look like uh real and they would pass through
32:52
the some of the softwares which may not be you know adapting uh adapting
32:58
quickly enough and um at the same time you have some still measures which could
33:04
detect these uh these patterns. You may have you know algorithm AI actually
33:08
which can track you know uh mouse movements uh you know login times um
33:15
even you know uh I mean you have a lot of elements which you can track but of
33:20
course once again you know uh regulated business is limited by what you can
33:25
collect what you can interact with or what consents you receive from the
33:28
customer and fra so of course we have seen that um and that doesn’t have to be
33:35
AI knowledge by the fraudsters. I mean there are tools ready made tools for a
33:39
lay person to to to to to start pulling pulling off uh you know fraudulent
33:45
schemes. Uh and that’s that’s that that’s what is really scary and of
33:50
course as an industry we have to be enabled to to to to to develop let’s say
33:55
country measures or even you know proactively uh do something about it.
34:01
Yeah, as you said Stephanie, I was exactly thinking the same. So fight fire
34:06
with fire. So fight technology with technology, right? This this is the case
34:11
with i gaming anti-fraud solutions all the way up to you know
34:17
serious situations that happen around the world. So if your technology is not
34:23
up to par with the aggressor’s technology then by default you will have
34:28
a deficit and therefore you either need to compensate for that deficit with
34:33
other means like more people more cost or invest in in the technology which
34:40
also has a cost on itself and needs special people to design test quality
34:46
assure and roll out and maintain uh programs. but also it’s much more
34:52
scalable on the long term. Plus, there’s one big thing that uh fraudsters have an
35:00
infinite selection of uh operators to target. But if they see that an operator
35:08
has very high guards, they will simply stop pushing and turn their efforts
35:16
where uh cases are much more easy to abuse and on top of that they will tell
35:22
their friends. So they will tell their network. So by setting high guards and
35:28
investing initially it’s not that you protect the business but you also set a
35:33
precedent you set a reputation that it’s not easy to um uh to uh to be abused. So
35:43
I think this is the meta the meta game is what is more important when it comes
35:48
to uh uh to to these waves of of attacks by you know from from from fraudsters
35:55
and obviously they will leverage AI in you know forging documents utilizing
36:02
bots and we have seen in the past a lot of scandals uh with bots usage
36:08
especially with poker clients and then companies had to on board very
36:14
specialized talent together with a lot of technology
36:19
to counter this. But then if you
36:22
>> if you act late, it becomes messy sometimes because you have to go back
36:28
analyze investigate and also stop the existing one. So um precaution always
36:36
precaution in this case
36:38
>> and I you know I I think about the visualization
36:41
uh you know because it is hard right as a business as you are trying to grow and
36:45
compete you do want to open the door you know and you do want to drive that
36:49
player acquisition and not put too many guard rails but at the same time going
36:54
back to who who you are to the fraudster I imagine you know everyone has traveled
37:00
a little bit maybe to some foreign countries that that you know maybe are
37:05
not the safest or have a high reputation for pickpocketing. I see that the speed
37:10
of AI adoption on the other side are making operators who are not ready like
37:15
a tourist going to a country waving their cash around. You you know you you
37:20
just don’t want to end up having that brand reputation where it’s easy. You
37:25
don’t want to be an easy target. And so Greg, you’ve helped me create this like
37:29
this very salient visualization about high guards. Do you want to be a a smart
37:34
tourist where you put a little bit investment in making sure your passport
37:38
is safe, you know, um versus being the operator who just goes to the the
37:44
country and has cash waving around. you know, you’re going to be a target and uh
37:49
and it’s going to be more and more apparent as you know, you mentioned the
37:54
the the sharing of the knowledge from the fraudsters and I think in compared
37:58
to industries when I look at other industries I gaming is particularly
38:03
um at risk because obviously the returns can be high but it’s also the age of
38:08
some of the um of eye gaming right um digit digit digitization of eye gaming
38:14
is bringing younger and younger uh users who are also at the same time
38:19
you’re welcoming welcoming new payment methods that don’t have the guardrails
38:23
like crypto. And so that breeds faster, smarter, digitally savvy um
38:32
fraudsters. And so more and more you’re going to look like the user who’s
38:35
flashing cash around. And so yes, the investment up front can be feel
38:42
daunting. Um, but it is more and more important that you’re not, you know, the
38:47
the easy one to target.
38:51
>> Very true. Very true, Stephanie. I think it’s uh
38:54
>> Yeah, I do think that’s really Yeah.
38:57
>> No, no, go, go, go, go, Joe. I mean, just uh
39:01
>> Oh, no, no. Yeah. Yeah. No, no, we are we are kind of uh edging towards the uh
39:05
the end of the session, but um I did want to uh touch on an important
39:09
discussion actually before we get there. um just around the cultural shift and
39:14
how we turn compliance into a business accelerator and how we can see uh how we
39:21
how we yeah how we change that perception that compliance is now uh or
39:25
should be a business accelerator rather than a hindrance. Uh P yeah kick us off
39:29
on that one. So yeah, I mean like I think in terms of fraud uh I mean what
39:34
is important is to let’s say make make uh the business aware of the metrics
39:39
connected that you know what is the what is the initial cost of on boarding the
39:42
customer and uh if you do it wrong basically what you lose uh in the long
39:49
term it’s like you know there’s a difference between dating and a marriage
39:53
as well if you have a successful marriage you know it lasts a long time
39:57
and even you know forever how how they say say it and uh and you can actually
40:04
you know mutually benefit uh between between yourselves. So it’s the same
40:09
with operators you know you you know that strategies of you know uh making
40:15
high profitability and you know very easily churn of the customers I mean
40:20
these are of the past. So I think this is like a cultural shift also brought
40:24
about by the regulations but also industry standards you know um
40:28
especially in the regulated markets where you really want to develop the
40:32
relationship with the customer you know uh on a sustainable long-term basis to
40:36
to create a sustainable business and um I think this is uh you know what you can
40:43
say as you know uh commercial uplift in the end reliance on on on let’s say such
40:50
a segment of players uh long term. where you can actually plan your business
40:54
objectives. uh it’s slightly more difficult to present because I think
40:58
it’s not uh also developed for AML let’s say so antimandering which which of
41:03
course is closely related to to to fraud but the same you know if you demonstrate
41:10
uh you know the savings and basically what is the initial cost and then if you
41:15
get things wrong uh is a cost wasted basically and versus you know
41:20
maintaining uh the relationship you know the proper tools the initial investment
41:25
and the returns over a couple of years time then you can you you can really
41:30
demonstrate that to them to the to the business. So of course uh then the
41:34
operators can compete uh based on you know what kind of solutions what kind of
41:40
tools are employed to comply with this requirements and how to establish this
41:44
relationship. So this translates into the user experience. Of course if the
41:48
regulator is prescriptive you it has to be ABC. it has to be displayed like that
41:53
say less competition but you know most of the times you don’t have these things
41:56
defined in the regulation so you can like Stephanie mentioned at some point
42:01
you can uh move the competition into designing you know the UAX and and and
42:06
and uh you know ease of of on boarding uh and the life cycle of the customer so
42:13
I think this could be a good good opportunity to you know implement within
42:18
let’s say wider wider compliance science framework. Um also this notion that uh
42:24
it can be you know a business enabler and competition uh enabler um uh not
42:30
just tick the box uh field basically not just the cost which you know you spend
42:35
for nothing. Yeah, I think Joe you made a fantastic
42:41
comment there and P covered most. If I can add there is that I feel there is a
42:47
a clear dichotomy of certain things that are non-negotiable like reporting
42:53
suspicion for AML reasons like there’s nothing to discuss there like it’s it’s
42:57
a it’s an obligation that protects the MLRO protects the company protects the
43:01
directors protects the reputation and then on there are other negotiable quote
43:07
unquote aspects where um we can get the compliance and the fraud or the AML or
43:15
the RG team into the business enabler status by applying risk management
43:21
statuses and offering options. This is what I tell my teams many times. You we
43:25
need to be offering options and different risk categories to the
43:30
decision makers so that everybody is on board with our approach and then
43:37
obviously re-evaluate a certain approach after a certain time frame. So we give a
43:43
sense of unity because you know if if a project goes on and there are three or
43:48
four different uh efforts pulling away it usually doesn’t end up well. But if
43:56
there is an alignment on the approach, everybody is aware of the risk,
44:00
everybody is aware of the pros and cons and what are the benefits and and the
44:06
risks, then the company moves as a as a unit as a whole and then this cultivates
44:12
culture because underlying that in terms of a compliance team being a business
44:19
enabler are you know many more you know relationship matters to be solved. So,
44:26
but then if if if there’s alignment on the goal and on an approach, I think
44:30
things move faster and more smoothly.
44:36
>> Yeah. And I I love the words that you use around unity and and culture shift.
44:40
You know, fraud and AI. It it sounds really, you know, fancy and um um not
44:48
something that’s tangible, but but at the end of the day, you know, um I
44:52
gaming is very much a human business, right? And historically, I’ve seen
44:57
payments here, RG here, fraud over here, and to your point, revenue over here,
45:03
and it it could be contentious. And this is why we always say there’s a balance
45:06
as if they’re opposite ends of the spectrum, right? Um, but Greg made a
45:11
really good point around unifying and understanding. I think a lot of it is
45:16
education around the culture shift is that, you know, the player doesn’t sit
45:20
in different departments. You know, they don’t think and compartmentalize that
45:24
way. The way that they act is is streamlined across their own journey.
45:30
And so, uh, the question was asked, how do you protect against AI? I think a lot
45:35
of the AI tools, you know, um when we rely heavily on just verification alone,
45:41
um you kind of miss the the behaviors that that happen uh you know, further
45:47
down the line. And so what what I say why I say that is because different
45:52
departments have different insights and different control mechanisms at
45:56
different parts of the player journey. And so if um a new policy or a trend
46:02
gets picked up by Greg’s team, you know, he has to be able to fast facilitate
46:07
that to the other departments because at they may have better insight and they
46:11
may have better control points earlier or later on in the journey to provide
46:16
that kind of cohesive um that cohesive fortified I think approach. And um one
46:24
of the things that I I really thought thought was important was the leaning
46:31
towards UX, right? Greg talked about the different optionalities. When you talk
46:36
about different optionalities in a unified front across the different
46:40
departments, all those departments become growth drivers, right? They talk
46:46
they all have a say in what level of friction. And I think we did have an
46:52
example um around um and I I I can’t speak it to I’m gonna pass it to Greg
46:58
Ptor about you know how why those risk signals um an example of those risk
47:03
signals and the speed of that communication is important. I wonder if
47:06
you guys can share an example that you shared with me earlier. I think it was
47:09
Greg that that you did that.
47:13
>> Yeah. Um, no I think it’s uh exactly what what you say Stefan is there and
47:19
and and and to add another extra layer of u let’s say let’s say difficulty 90%
47:28
of the customers are silent. So it’s it’s going to be you know depending on
47:35
on on the on the on on the approach that the company says the if if if we don’t
47:41
get it right then the impact will be really uh quick in terms of alignment in
47:48
terms of let’s say applying a framework of protection for RG for fraud for
47:54
especially for fraud where we need to keep the database clean as much as
47:59
possible. So then you can enable marketing and CRM to focus on their uh
48:05
on their let’s say on their activities on on clean databases and and put the
48:11
marketing budget where it needs to be not reiving bonuses to previous bonus
48:17
abusers or uh you know uh so that’s that that’s a really important aspect.
48:26
>> Yeah. Interesting. And then um Pa did you want did you want to come in there
48:30
with a with one as well?
48:32
>> Exactly. I mean I can only concur here. I mean also you know like if you have
48:36
basically compartmentalization and silos I think this is also important to break
48:41
up the silos uh you know like with the tools uh you may have uh you know money
48:48
funds misallocated between uh you know the goals as Greg mentioned you know
48:52
marketing commercial you know bonus abusers. Why I mean certain behaviors
48:57
could have been detected escalated between the teams and uh you know
49:02
there’s a need for you know uh unified maybe not unified but let’s say inter
49:07
operate operative tools which are accessible by different uh departments
49:12
so like common alerts you know uh notifications and so on so so so I would
49:17
say this is a great example you know like I think which is pretty common in
49:20
the market
49:24
>> we’ve had an extensive discussion but I’m glad we have highlighted that that
49:29
uh breaking down silos within companies is so crucial because especially as
49:33
companies scale up as well because that’s um a really impactful tool when
49:37
it comes to uh tackling fraud and uh eradicating fraudsters and detecting
49:42
fraudsters as well. Um we we do have a little bit of a closer that’s come in
49:46
from our our audience and Greg you mentioned poker before obviously a a
49:50
very a very big one. Um but just uh yeah in one word in I guess a brief
49:57
description um is there an area of I gaming whether it’s sports book or
50:02
casino or specific games or something like that that specifically more
50:07
frequently fraudsters utilize AI within that area. Is there is there one area
50:12
that really is a hot bed for frauds to AI activity? Obviously, we’ve mentioned
50:16
poker uh you one there, but any others? Um Steph, kick us off.
50:22
>> Yeah, uh you know, sports betting I I I think is is a ground for um for a lot of
50:28
fraudsters. It’s again I think the demographic of sports betting is um
50:34
slightly younger and so that drives a lot of that and the opportunity to game
50:39
systems um you know I see a rise in you know different betting mechanisms um
50:46
that they’re trying to take advantage of um but the digitization of sports
50:52
betting I think breeds a lot of um a lot of fraud so the most interesting ones I
50:58
think too as Oh,
51:01
>> interesting.
51:01
>> No, the f between the operators I mean not I mean it’s just to be clear you
51:07
know it’s very difficult to tamper with let’s say the systems the game systems
51:12
it themselves it’s more about you know like exploring what’s happens in the
51:16
background and and so on. So as Stephanie mentioned sports betting you
51:20
can have uh you know I mean players can have accounts on different uh platforms
51:24
I mean with different operators and of course the speed of information sharing
51:28
and the speed of uh you know interactions between the frauding can
51:32
can actually you know make it uh more attractive uh to use you know the
51:38
leverage of and different of difference of controls between different operators
51:41
you know on the same market. Let’s say if you have sports betting and you know
51:45
uh abusing you know the odds or uh trying to to to to to I mean to take
51:51
advantage of this maybe way easier with AI which enhances the speed of decisions
51:55
and speed of enabling information.
51:58
>> Yeah I think Joe I think AI is in in many more categories than we can think
52:05
actually. So it’s there in KYC forging documents is what Stephanie said in
52:11
sports book and you know exploiting exploiting angles that sports books and
52:20
um odds providers cannot really get there. Obviously, we see the the poker.
52:27
We we didn’t even touch on the agentic AI aspects of, you know, creating farms
52:33
of accounts just playing and playing day in day out and exploding exploiting
52:40
bonuses and casinos. So, it’s it’s it’s really a big big wave that is coming.
52:46
So, we better uh get get the protections in place.
52:51
>> Yeah. Yeah. a warning that really um underpins the importance of the
52:55
discussion that we have just had and uh yeah the importance of what you do as
52:59
well. So um I thank you all for your time today. Uh it’s been really
53:03
insightful. Uh lots of really important nuggets to take away. I’m going to have
53:08
to listen back and uh yeah pick through them and make some notes. So uh yeah,
53:13
thanks ever so much and thank you everybody for listening. Thank you to
53:16
everybody that engaged and asked questions as well. um appreciate your
53:20
time and yeah, I look forward to speaking again soon.
53:23
>> Yeah. Thank you.