From Login to Loyalty: Using AI to Build Trust and Block Fraud

0:00
Excellent. So, hey everybody. Super excited to talk to you all. Uh again, my
0:06
name is JP Bllejo. I’m the head of product marketing at SIFT. And uh my
0:11
work here focuses on helping businesses understand the benefits around AI uh
0:19
when it comes to protecting their consumers from fraud. Um and it’s also a
0:24
great discussion uh or this will be a great discussion as I talk about how not
0:29
just protecting consumers from fraud is key but how
0:34
using AI can be used as an opportunity or a growth model for your organization.
0:41
So through the session I’m going to share how AIdriven identity and risk
0:46
decisioning can stop attacks before they start. we’re moving into being a much
0:51
more um proactive uh type of solution uh than we have been
0:56
in the past. And more importantly, I want to show you how trust when done
1:01
right does actually become a growth engine. So we all know because we are
1:07
consumers and customers as well as vendors and um organizations that
1:13
support consumers. We all know that speed, convenience and safety are
1:18
expectations. Uh and there is no difference between
1:22
the three of them. Meaning that consumers want a seamless experience. Uh
1:27
they they view that the risk is the burden of the vendor itself, not their
1:32
own. Um and that’s a tough balance uh because the minute you as the vendor or
1:39
as the merchant add friction, these consumers tend to to leave. Um and then
1:45
the moment you loosen those defenses and creating a more seamless experience for
1:50
consumers, fraudsters find a way into those weaknesses or into those loosened
1:56
uh areas and uh infiltrate. So, I’ll walk you through how AIdriven identity
2:02
and risk decision are changing that equation. Um, helping helping to block
2:07
fraud before it starts while giving legitimate customers the seamless
2:12
experience that they demand. And ultimately, it’s about showing that
2:16
trust isn’t just a defensive strategy. It’s a driver of revenue, loyalty, and
2:22
that long-term growth. But fraud tactics continue to evolve.
2:27
It’s, you know, a never- ending cycle. I think those of us who have worked in the
2:31
cyber security industry have known that it’s a self-fulfilling prophecy. We
2:36
create defenses to protect consumers against attack as fraudsters and cyber
2:42
criminals continue to evolve and create new ways of infiltrating networks and
2:47
systems and applications and customer identities. So, it is this vicious
2:52
cycle. Um but I do think that AI is a gamecher for us for us as um an industry
3:00
being the the fra fraud vendor but also uh for you all to adopt and actually
3:06
start worrying a little less around fraud and focus more on growth. So one
3:12
of the biggest shifts that we have seen is that fraudsters are moving away from
3:20
focusing on the login aspect of the consumer journey and looking for those
3:25
weaknesses that go across the entire consumer journey. And one of the stats
3:29
that we have seen is on this very simplistic uh life cycle that I have
3:34
here the post payment section we’re seeing a significant increase around
3:40
policy abuse. um or an opportunity for those fraudsters to um enter an um the
3:48
customer’s journey, take ownership of that identity and then
3:52
take advantage of that in future sessions and interactions. But there are
3:56
some other areas that we do see increased risk. One is around credential
4:01
stuff stuffing where attackers use stolen user and password combinations at
4:06
scale often with automation to gain access into those accounts. Um, and the
4:11
real damage doesn’t actually happen at login. It happens after because once
4:14
inside processors exploit stored payment methods, the those loyalty balances and
4:20
also still the customer data within those accounts. So no matter where a
4:25
consumer is in their journey, risk continues to be a factor. So, I’m going
4:31
to talk from both ends of the the spectrum saying, you know, we need to
4:36
lessen restrictions while at the same time increas increasing defenses. Um,
4:41
we’ve also uh seen a rise in synthetic identities. And I think that this is a
4:45
really interesting uh risk vector because it stitches together real and
4:52
fake information. So, it may take a social security number of a legitimate
4:57
individual and change the name. Um, may use a real address or a real credit card
5:04
number and then change the name or have a fake address. All of these
5:08
combinations make it much more challenging to uh identify whether that
5:13
is a fraudulent individual, fraudulent account, etc. And then of course,
5:18
account takeover which will be filtered throughout my conversation here. um that
5:23
remains to be a top threat. So at the end of the day, everything is of serious
5:27
concern. But what connects all of those different fraud tactics together is that
5:31
they target the identity as the weakest point, the individual themselves. And I
5:38
don’t necessarily think that that is by nature a gamecher.
5:43
Where I think the bigger risk and where the bigger concerns are is that we as
5:48
individuals have dozens if not unfortunately hundreds of unique um
5:55
identities in this digital world. Every account that we open up and create,
6:00
every password uh account that we have is a unique digital identity. So even
6:06
though at the end of the day you’re trying to protect JP Bllejo on your
6:08
systems JP Bllejo sits in many locations under many accounts on many websites uh
6:15
so it becomes you know a really challenge really huge challenge to
6:21
protect the identity of an individual. So AI really is transforming the fraud
6:28
tech landscape from both extremes. So I’ll not to throw too much fear or FUD
6:33
into the conversation, but where you know some of our anti-colagues, the
6:39
fraudsters and cyber criminals are taking advantage of it is they’re using
6:42
AI to create these virtual identities as well as attack um good identities out
6:50
there through different forms of fraud. And I’ve listed a few of them here and
6:54
three that I’m going to f I’m going to just touch on because I think that
6:57
they’re very fascinating and terrifying at the same time is one is on improving
7:02
written language. So as a marketeteer I use AI a lot to help refine the message
7:09
get to the core discussion topic that I want on content when I’m creating it or
7:14
writing a blog etc. But it also helps me make sure that I’m using the correct
7:18
grammar. And for cyber criminals, it allows them to align their efforts uh to
7:26
match the language of uh or the of the identity and of the nationality of the
7:31
individual. So, let’s say, you know, I’m US-based and I’m a fraudster and I’m
7:35
trying to uh attack or at least uh create a fake identity on someone from
7:43
France, I can use AI in translating into a local language, which then makes it
7:49
even less obvious that I may be the fraudster whenever I’m accessing an
7:53
account or communicating with uh an organization and trying to dispute a
7:58
return. Um so I think improve written language works on both sides of the
8:02
spectrum. The other one is on the um I’ll call that generating malicious
8:07
code. Unfortunately AI can make anybody a tech expert these days. And so even
8:14
though 10 years ago I would have a very similar presentation where I would talk
8:18
about YouTube training individuals now you can actually use AI to generate
8:23
malicious code in real time and inject that in real time. Uh and then the other
8:29
one is creating visual deep uh and and creating those deep fakes. Um this sits
8:35
as a concern especi especially in the fintech and the financial institutions
8:40
uh where biometrics is being adopted more and more where you’re required to
8:45
either hold up a form a picture ID or they require some type of visual
8:50
screenshot of you where they’re able to use deep fake technologies to create a
8:55
three-dimensional image of yourself and front that in front of the camera
8:59
itself. So AI is transforming the fraud attack landscape, but in defense, it’s
9:07
also helping us accelerate our defense strategies in being able to identify
9:13
those anomalies in real time and also making advancements without having to go
9:19
through multiple validation steps and processes that we have kind of built up
9:24
in our own defense mechanisms. So the broken trade-off really comes to
9:30
embracing AI as part of your fraud strategy and your decisioning engine uh
9:36
and you have to overcome some of the defense strategies that you use you have
9:41
used before. So the old approach, more rules, more friction. I I can’t
9:46
emphasize enough that when using rules in your
9:51
defenses, uh your fraud defenses, it becomes a challenge because the more
9:56
rules you create, the more challenging it is to go through the journey, meaning
10:01
it more friction is being added to the consumer journey. And then when you make
10:05
a change to one rule, you have to go back and see does it impact any of the
10:09
other rules uh that you may have. And I do have one uh example of a customer uh
10:16
that prior to adopting an AI based fraud platform solution, they had more than
10:22
300 rules set up in their environment. Um, and by adopting the technology, and
10:28
this isn’t a cell, this is just a stat, that they are now down to less than 25
10:32
rules and continuously being able to take a look at those rules
10:36
and reduce them over time because the platform that that they’re using from
10:41
SIFT is becoming smarter and is understanding the the different use
10:45
cases around their customers and being able to eliminate more and more rules
10:50
because it’s looking at each individual and each identity in real time and
10:54
analyzing it against the greater population and that outcome becomes less
11:00
blocked genuine behaviors, more fraud being blocked and then
11:06
creating that opportunity for consumers to grow more trust into the experience
11:11
that they have with you and growing the revenue whether that’s through more
11:15
purchases or larger purchases. Um because those customers expectations are
11:20
built upon experiences that they have whether it’s on your site or on your app
11:26
um or someone else’s. They they do benchmarking and baselining and they
11:30
determine which ones are the better performers and they set their priorities
11:34
over there. So the challenge isn’t just fighting fraud. It’s doing it without
11:38
slowing down the very customers you’re trying to serve.
11:44
Now, couple areas where AI does have that true opportunity of really
11:49
accelerating is I’ve I’ve listed here I kind of stolen a a marketing slide here
11:54
um to kind of define what a fraud platform should look like. And what I
11:59
put here is I’ve highlighted six of the key I’ll say processes that a fraud
12:05
analyst needs um to analyze or needs to participate in to assure that fraud
12:12
isn’t occurring. Um so that rules and policy writing um investigations
12:17
looking at risk scores and the factors associated with those risk scores. Um
12:24
all of these come into play but the reality is when you’re looking as an
12:27
individual who has to either be proactive or reactive to a complaint.
12:33
All of these uh areas take time and adopting an AI powered identity and risk
12:40
decisioning platform allows the AI components to look at all of these
12:47
at the same time simultaneously in real time so that you’re not having
12:52
to prioritize. You’re not having to look at the customer journey and identify
12:58
patterns that the hum that the that the individual has had. The AI is doing all
13:03
of this in real time for you and informing you as the fraud an uh analyst
13:07
where there may be risk along that consumer journey. And those can really
13:13
be anything from uh a slowdown in how you are making the
13:19
purchase to abandoning a cart which is atypical of an individual to entering a
13:25
new credit card. Uh all of those may be anomalies that you as a vendor may need
13:32
to that you may want to create any uh some form of step up in uh
13:37
authentication requirements. But it’s empowering you with all the information
13:41
so that you can make real-time decisions so that you don’t have to do post
13:45
investigation management. So an identity first practice like at
13:53
the core and this is a little bit about sift and um who we are and how we have
13:59
built our AI fraud platform at the core of our platform is an identity first
14:04
approach um instead of looking at transactions and loginins and iso is
14:09
isolation we unify those signals across the entire customer journey. So that’s
14:15
login, device elements, payments, account information,
14:21
behavioral biometrics. We take a look at all of those signals across the the
14:26
entire customer journey um and do the analysis because we do know that
14:31
behaviors do change. We do know that uh customers may adopt a different use case
14:37
over time. But when we have full visibility into the consumer journey, we
14:42
can determine if any of those anomalies are one-time, are risky behaviors that
14:48
don’t necessarily map to the nuances of the individual or of the account, or if
14:54
they’re criminal by nature, and put policies in place to to allow to go
14:59
through to inject a step up uh or just deny access or deny the transaction from
15:06
from happening. Um, so this framework allows us to leverage consortium data.
15:11
So as Ed mentioned very early in the presentation, JP Blleo probably has
15:15
hundreds of accounts or hundreds of digital identities out there. Um, by
15:19
leveraging our consortium data, we get visibility into who JP actually is from
15:25
all of those different accounts. uh and then allow the individual or allow the
15:31
uh you as the vendor to make decisions on whether JP in your environment is
15:38
really JP or not. So we create pre-built workflows that let teams deploy defenses
15:44
um in seconds as opposed to months. And because our model continuously learns,
15:50
the protection doesn’t stop at login. The analysis doesn’t stop at login. It
15:55
continues to analyze behavioral behaviors, transactions, movements,
16:00
non-movements through the customer journey. Um, so we get smarter uh as
16:06
your consumers use your platforms, your tools, your
16:10
applications. Um, and this is where the game changes for AI from the perspective
16:16
of legitimate vendors against legitimate consumers. Um, as Simon had mentioned,
16:23
we are powered by more than one trillion digital interactions every year. So, our
16:27
global identity network surfaces patterns across multiple industries,
16:31
multiple geographies, um, and multiple attack vectors. And real-time behavioral
16:37
signals and machine learning allow us to dynamically decide the risk. um based on
16:43
what we see globally in your in a specific industry or in a specific
16:48
region could ultimately raise the risk factor of what’s happening in your
16:52
environment at that moment or lower the risk at that point in time. Uh so that
16:58
global visibility um allows us to look at every individual through multiple
17:03
lenses and guide you as as the vendor or the uh the customer to make decisions on
17:10
your own consumers allowing identity to become much more of a secured measure
17:16
without adding those layers of friction that have traditionally been part of a
17:21
fraud platform such as the slide I mentioned before.
17:26
So when we talk about the impact that it has to individuals and how an AI fraud
17:33
platform does speed up the process of defending, some of the benefits that you
17:39
can see um is the reduction in chargebacks,
17:43
the increased approval rates that would happen, preventing bonus abuse, um
17:49
especially in industries like I gaming and fintech um where we help to ensure
17:54
that frauders can’t in exploit those incentives uh or drain the accounts that
17:59
are h that that they have. Um it allows to protect those deposits and then also
18:05
stop promotional abuse whether it’s from a friendly customer or whether it’s from
18:10
a fraudster itself. It allows you to look at JP again in your environment to
18:16
see if he’s created multiple accounts and is trying to get that firsttime
18:20
buyer discount or not. And ultimately all of these protections allow you to
18:26
save margins and grow your business increasing that capability or that trust
18:32
within the organ within your consumer base. So what should you take away from
18:37
my very brief presentation? Um first that closing the identity and
18:42
verification gaps with real time intelligence by adopting AI uh
18:48
technologies um has to be your starting point. It
18:51
doesn’t mean you have to embrace it all at once. Really focus on understanding
18:56
your consumer’s journey. Look at all the steps in which they go through where
19:03
data uh sits on the different web pages or on the different applications, how
19:08
they interact with that, identify the weaknesses and start there because
19:12
attackers are cons consistently probing for those particular weak spots. So
19:18
whether that’s at login, at checkout, or somewhere in between,
19:22
um start at that point. Second, if you have to deliver trust without friction,
19:27
meaning you’re talking milliseconds of delay will stop a customer from moving
19:32
forward, um every extra verification step, every false decline, every
19:38
unnecessary password reset chips away at that particular brand loyalty. So AI can
19:43
help streamline verification. um whether it’s a brand new customer or
19:48
an existing customer just logging in. Um it helps keep those accounts safe uh and
19:54
then allow for those repeat businesses and higher approval rates. And finally,
19:59
fraud prevention needs to be seen as more of loss avoidance. Um, it’s a
20:05
growth lever and protecting accounts and experiences translate into customer
20:10
retention, repeat purchases, and long-term loyalty because customers when
20:16
they know they’re safe come back. It’s as simple as that.
20:21
So the opportunity in front of us is to
20:25
change the way we think about trust and the way we think about identities
20:30
because too often business is treated as a cost of doing business and we have to
20:35
change that u because in reality trust should be a competitive differentiator
20:40
or a competitive advantage for you as the business. Trust is what determines
20:45
whether people complete the purchase, come back again, or ultimately stick
20:49
with you as the brand preference because customers that embrace AI powered trust
20:55
frameworks aren’t just keeping pace with the attackers. You’re becoming more
21:00
proactive and and being empowered to stop fraud before it actually happens
21:06
and close that gap into the weaknesses across the consumer journey. And that’s
21:11
kind of exactly where CIF sits and that’s where we we place a lot of our
21:15
intelligence and our expertise in that area. That’s what excites us as an
21:20
organization every day. And that’s how we collaborate with organizations like
21:23
you to better understand your consumer’s experience, your consumer’s uh kind of
21:29
process in which they go through their journey and help you build the defenses
21:33
that are necessary to be proactive to allow for growth to occur. So as you
21:38
leave today, I’d encourage you to rethink trust not just as a defensive
21:41
shield, but as the foundation of lasting customer relation relationships and
21:47
competitive differentiation. The businesses that invest in it now will be
21:51
the ones that grow and thrive in the digital economy to to come. And so with
21:56
that, I would like to thank you for the time. I hope I didn’t take up too much.
22:00
I know I’ve covered a lot and bounced around everywhere across the map here.
22:04
uh definitely open to taking questions and furthering the conversation offline
22:08
if necessary. So, Simon, are there any questions?
22:12
>> Yeah, RJP, thank you so much firstly for your uh really uh great informative
22:15
presentation. We do appreciate it. Uh two quick questions from the floor. Uh
22:20
what role does AI play in balancing fraud prevention with a seamless
22:25
customer experience?
22:28
>> So, I kind of peppered it into the presentation
22:32
um in there. So, I’ll use I’ll just give you a quick example on on where AI does
22:38
look into it. So, let’s say I’m logging into one of your websites as a new
22:43
customer and I give you some personal information, you know, to create my
22:48
name, my account name, and my password. That’s one step check where you can look
22:53
at the device that’s being logged in or the accounts being created. than using
22:58
our global network. You can see if that device has been used in any other
23:03
accounts and if another vendor has identified that that device or my device
23:08
has been used for fraudulent purposes, you’re alerted to that at the point of
23:11
account creation. That’s just one small example and where the consortium data
23:16
and the AI can go out and actively search while I’m still typing in
23:20
information or shopping in the account.
23:24
>> Perfect. Thank you so much indeed. And finally, what’s the biggest challenge
23:30
with AI in security?
23:32
>> Adoption. I I and I going to use a marketing phrase on that. The biggest
23:37
challenge is acceptance and adoption. I think people
23:42
we are I’ll say to a certain extent AI as a collective is in the earlier stages
23:47
or maybe we’ve accepted it um and now it’s in that explosive growth mode. But
23:52
the reality is machine learning and which has now evolved into AI has been
23:57
around for a while and vendors like uh SIFT have for 10 plus years been focused
24:04
on making sure that as this evolves as an entity that we are looking at that
24:08
evolution assuring that it’s being used for the
24:12
right purposes uh and then being deployed in real time. So again the
24:17
emphasis is on acceptance just embracing the fact that this is how we have to do
24:21
business and asking the right and questions meaning asking vendors who
24:26
embrace it as to how and how can you use it and how can you adopt it and being
24:30
open-minded.
24:32
>> Thank you so much indeed for answering the questions and for the questions from
24:35
the floor and also once again uh JP thank you for your presentation as well.
24:39
We do appreciate you giving up your time. I know you’re going to join us on
24:41
the panel uh a little bit later but for now thank you so much indeed.
24:46
>> Thank you Simon and thank you everybody.
24:49
>> More than welcome. Thank you once again to Jean Paul there the senior director
24:53
product marketing and of course our platinum sponsor SIF discussing from
24:56
login to loyalty using AI to build trust and block fraud. All right when we come
25:01
back we’re going to heard here from our third and final keynote of today we’re
25:04
going to be discussing AI powered identity building trust and security in
25:08
a zero trust world. uh with dipst and that’s going to come very very shortly
25:14
after this short break. Heat.
25:47
Heat.