Who is Liable in Agentic Payments? The Question the Industry Can't Answer Yet | Camilla Bullock, CEO of Emerging Payments Association Asia

Camilla: [: 00:00:00

Monica: we're in Money 20/20, the industry is changing super fast. The theme that we have is agentic payments. So there's a stat that says across APAC, 81% of customers are e- expect to shop with agentic AI.

81% of customers, that's a lot. And this shift could drive over 1 trillion in spending. Wah. So it seems like there's a big promise out there of what agentic AI could do, but at the same time, there's this big question about fraud, liability, quantum security, like there's so many questions

so [: 00:01:00

we have that audit trail if [: 00:02:00

Or I would say I hope we don't reach 85% before we have those things right because the industry, and I, I guess even society, has so much to lose because once the trust is gone, uh, it's gone, and it will take a very long time to build up again. So It's exciting, but now is the time to really to come together to unlock this opportunity because it has so much upside, but it could also have a lot of downside.

the same. Hmm. It's like the [: 00:03:00

We need more governance audit trail so that we can innovate, which is interesting.

Camilla: Yeah. It's... I think a word that's come up and that I'm pondering back and forwards is that we always seen, uh, agentic AI or the LLMs and that to be a little bit of black box that we don't know how, how it's behaving, and that's why we're a little bit s- scared, like, are we doing the right thing?

If the regu- regulator is knocking on our door, can we explain, explainability how we came to decisions? But I would like to move the whole industry to talk about positive black boxing like you have in the aviation industry. Because if when something goes wrong, then you go to the black box- You go to the black box

that's where the audit trail sits. Oh,

Monica: cool So

e thing to shift because the [: 00:04:00

I've been in so many... Well, like not so many, but I've been in four round tables within the last two months, and I do find the thought process is- Evolved ... is actually matured. And, um, that know your agent or KYA. I said it right now. Yeah. KYA is something like we have to solve now because many participants w- will probably say no to allow agents, uh, to execute.

e of that, um, uh, purchase. [: 00:05:00

Monica: Curiosity to question, who, who's deciding that? You know? Like, is it the schemes? Like, is it the big banks on behalf of the industry?

Like, who's making those decisions?

Camilla: Uh, so I think it's just so early now, so anybody that is risk-averse would probably say, "No, we're not going to let this ex- execute go through before we know the liability," because- It, it is a big risk there because it's not just one purchase or one trade- No, exactly.

Yeah ... it can be many, and it can be big amounts. Uh-

Monica: That's really unworthing.

Camilla: Exactly. So, so many of those parameters are not in place yet, so hence I'm a, a bit bearish on the 85% in a quick pace, but it's definitely a possibility. And I think also what the industry needs to really look at for what use cases does it make sense to start this?

What use [: 00:06:00

There's all those things- Facts ... we actually haven't defined, and, uh, the example I hear a lot like, "Oh, wouldn't it be great to have an agent to, to buy, uh, airline tickets?" But buying airline tickets for me as an individual is very complex.

Monica: Complex. Yeah, it is. '

Camilla: Cause it like-

Monica: Especially crossing to the other side of the world

Camilla: what time of the day are you flying? What route? Is your best friend actually in Delhi when you're flying there or not? It's going to change everything, but if it's a business that actually purchase a airline tickets might be slightly simpler because they're going to ask you certain parameters, and you can't choose.

So I think [: 00:07:00

But as an individual, as a business, when is the ips- upside big enough that you're happy for the agent to do it by themself instead of the discovery? Everybody has done research today, but do we want to take that extra bit of actually the money's leaving your bank account?

Monica: I, I think you raise, like, a very good point because it's like money will leave bank accounts, and all is good when it's the happy path, and that's it.

be wrong amount, wrong, uh, [: 00:08:00

And

Camilla: then also, uh, I hosted a roundtable in Japan earlier this year. It was one ... the first discussion I had designed on this topic, and it was interesting to hear everybody had so different views to the point that myself as moderators was second-guessing myself through the whole roundta- You've been in those, haven't you?

uld all be built differently.[: 00:09:00

So is it like that, know your agent, and is it a certificate on it, like this one has actually been tested, that it will stay within the guardrails. It's, it's, yeah, so many new things, you know?

Monica: Yeah, that's a very good point because then it's like I, as a non-coder with the technology, I can go and build an agent, right?

But the quality of my agent may not be the quality of the agent provided by a merchant or by a, you know, like an established institution. So there's going to be different, uh, reliability, quality, security of the agents that we're using.

Camilla: Yeah, you're noticing- And

Monica: that quality could- ...

Camilla: on Reddit today, uh, a- and I know that you way more advanced in using this than me- No, I'm not

ays. So it's almost like you [: 00:10:00

What would you buy? What would you let an agent buy, and for how, how much does the upside need to be for you to take the risk?

Monica: Right now, I think I could buy small, small purchases. You know the ones that you remember, oh, I need to buy, I don't know, a new, I don't know, like something for day to day, and I, I speak with Claude a lot.

The

Camilla: garbage bags?

Monica: Yeah, that's what I thought. But I thought about that, but I'm like, no, that's not a good example. But yeah, the garbage bags. You're like, "Ah, yeah, I'm running out of garbage bags," and I would love to be like, "Hey, Claude. Please can you get me, uh, garbage bags delivered to the house? Go to my c- go to my shop here, Lazada- Mm

ave, I would be happy to do. [: 00:11:00

Camilla: Yeah.

Monica: And then it's like, that's it. I would not use it with my credit card.

And not for, not for fights. Yeah. Like... But it's, like, those little things that you're like, "Oh, I need this extra little thing," just go to Shopee. The

Camilla: boring day-to-day, and the upside is actually not the money, it's the time you save.

Monica: It's the time. Yeah. It's the convenience. Because usually you have those thoughts, "Oh, I need to buy that," in the most random moment.

Camilla: Yeah. And

Monica: you don't buy it there. You're like- Yeah ...

Camilla: it comes with you or- So you... But you're almost thinking more thought execution, that it's when you hit you, you want to speak and say- Yes ... so the authorization is not going to be for five months until it finds the perfect garbage bags for you.

Monica: No, it's like a-

Camilla: Just- Yeah

in... This was actually not [: 00:12:00

Because when Taylor Swift releases her tickets to KL, they want to buy the, uh, hotels there and then. And that might be other things you would say, like, when is the most popular artist going to go on tour? You know, start to do those different complex things to find, uh... What did you say? I was gonna say, but you know, d- d- opportunities for upside.

Monica: Yeah.

Camilla: So is it opportunity for upside, or is it saving you time? So it's like, what parameters will drive the use cases? I still think it will be business use cases because it might much easier to put the parameters around it.

Monica: Hmm. That makes sense.

Camilla: Businesses aren't as impulsive as-

Monica: No, as- ... as you and me ... as, like, yeah, I need I...

Oh, no, [: 00:13:00

But if I just go, "Claude- Yeah ... can you please buy this? Oh, yeah, yeah. Can you buy that? Can you buy that? Can you buy that?" Because it's so easy, it removes friction, that people may start, or I could start overspending in stuff that I don't need. Yeah. It may be cheap, but it's like-

Camilla: Yeah. Yeah, yeah. No, no, it's interesting It's- Like, I,

Monica: I don't know Or the, there's a sale in...

My, m- my email is full, full of sales, like, every day. I,

Camilla: I might save money on this because I, I think you know, I told you that, like, I'm a big vintage shopper-

Monica: I just,

ng for, and I always have to [: 00:14:00

Uh, but yeah, those are those... It's edge use cases, and that's not... If I would say one thing to the industry out there today, it is don't look for the don't solve for the edge use cases. What is it that's really going to be worth it? Where is the upside big enough for the potential risk?

Monica: I like that. It's going to go in a clip.

omptPay. What lessons can we [: 00:15:00

Camilla: First, I mean, we've been very innovative is- in this region.

Um, I'm so proud of working in payments in our region wh- when I travel, and I speak about the use cases solving for very niches and, and challenges we had. I do think we're very early. I have had to, uh, realize that with real-time payments comes real time, uh, fraud and scam, so we already see a lot of good collaborations around that.

with that also we have seen [: 00:16:00

Like MAS has done, uh, a lot of good works around, uh, AI frameworks and guardrails- Mm ... and principles. What I do think a- and this is in general, we're coming to a little bit... You know sometimes when you have discussion in, in different location and it comes a really catchy saying, and that is like, "Oh, it has to be shared liability"?

That is not an answer because we don't know how to share it today.

Monica: That's what I was going to say- Yes ... I'm like, I'm not sure about shared liability.

Camilla: Yeah, we have to define then who is even in the value chain. Uh, so a lot of the challenges we have will also touch on, on fraud and scams, and how can we use data today because signals is going to be a very important thing.

k with data and how we share [: 00:17:00

Monica: Hmm. That's a very good point. And I want to go back to what you said about cross-border payments. It's amazing. You know, like the region, it's amazing real-time payments, cross-border real-time payments with all the rails, but it also comes with, uh, higher fraud, and then it's, uh, yeah, it's cross-border fraud.

You empty the account and that's

Camilla: it. Mm.

Monica: So I was just... As, as you were speaking, I was thinking if we put ourselves in the shoes of the fraudsters, agentic payments, it's just going to make their lives easier. So much

Camilla: easier, yeah.

Monica: They'll be like, "Yay, I can do this faster."

Camilla: And they have much more budget than- A lot

n is one of the biggest ways [: 00:18:00

So I think how we think around fraud and scams and liability will change because of AI making it so much more powerful, so you can't talk about neglect in the same way as we've done with, um, with the princes of, uh, scamming and so on and so forth. But then if you put the agents in there, there's like these agents on hyper being able- Exactly

to, as an industry together, [: 00:19:00

I mean, we are a network, so we're already working together, but also we're just growing industry. We're still growing double digits, so you don't have to be like, "I'm not going to work with that organization or that organization." It's really like, let's do it together and, and grow that pie because everybody's get a bigger slice.

Monica: Yeah, and I, I think that's another theme that I heard. Exactly, working together. It's n- it's, uh ... Yeah, we'll just have to work together to figure it out, work together with regulators, and I think, like you said, it's like trust. I think the conversation so far has been how do I, as a individual organization, build trust with my customers?

I think now it's the time of [: 00:20:00

Camilla: Yeah.

Monica: That's it. It's like implied trust.

You're like, "Okay, I can pay here." Yeah. That's it. Done. But it's like now as we introduce agentic payments, it's like customers are going to be ... It's not if we have, similar to crypto, if we have one n- not bad actor, but like one bank that messes it up , it will go viral.

Camilla: Yeah.

Monica: And then trust is not going to be eroded for the bank's customers.

It's going to be eroded for industry and agentic payments.

Camilla: Yeah.

ld trust collectively? Yeah. [: 00:21:00

Camilla: And that's what, uh, uh, go back to what I said about data. So I've, I have a data background, so hence I get very passionate talking about data often. And, and not just data point, but insights from data points, and this is something we should double down on because there is no payments without the movement of data.

Payments is data. Uh- Yeah ... and how we can start to, to share data between entities in a more purposeful way. I mean, there's a lot of pilots on the way already. In Australia you have one of the banks sharing the data with one of the telecoms, and so they can actually track that signals and early flagging of that this looks like, uh, fraud.

So more of those initiatives, and I think there are those conjunction industry to ourself like telecom, uh, and big tech and platforms that have a lot to win if we can find a way how we collaborate.

e gears, talking about data. [: 00:22:00

Um, it's not just that we're seeing change in AI and agentic payments. There's another work stream or another uh, tech that's evolving that it's like quantum. So, uh, your research found that less than 20% of senior leaders in payments, uh, are actively aware or tracking the quantum threat. What does that mean?

Can you tell us, like, what's the threat? What's quantum? What's the issue?

Camilla: This is one of the scariest thing we're working with, and if I think too much about it, I'm like, "Why is everybody sleeping at the steering wheel?" And to the point that I've gone to people and said, "Is it just me, or is it as scary? Is this threat as big as I understand it to be?"

talking about is... And this [: 00:23:00

Uh, so that could mean that they could steal your data or get access to your accounts. But it can also be, and most likely would start as, uh, attacks on things that changes how you do business. For example, say, a sanction list. Some- somebody gets access to a sanction list and go and change the entities there, and they go through.

Oh, yeah. You say, so people, you think of this like, "Oh, somebody's going to take my data as a private." What if somebody hacks and get keys to get into the liquidity of a bank?

Monica: Oh.

Camilla: Yeah, [: 00:24:00

is real." Google put up, uh,: 2029

's so many participants, and [: 00:25:00

Did I scare you now?

Monica: I'm like, I had a conversation just a day before coming here-

Camilla: Yeah ...

Monica: with someone that works in quantum.

Camilla: Oh,

Monica: nice. And, and he was, like, saying exactly like you. He's like, "People are, like not really taking this seriously." No. It's a big threat. And even I'm like, like, "What do you mean?" It's massive.

I think there's lack of awareness.

Camilla: And I think the challenge we're at is it's so many migrations going on already. It's so many technology, uh, uh, so many technology questions on boards' agendas today because this is a board decision. Yeah. It's not the somebody in the tech department. No. It's a board decision.

problem," while we, like, do [: 00:26:00

Seven

Monica: years?

Camilla: It's like, look how long it took for ISO migration, and some are at it still, so

Monica: Okay.

Camilla: Should we finish on a happy note?

Monica: Yeah. I'm like, yeah, let's go back to, like, happy, happy, happy, uh not happy, but it's like, let's move away from risks and threats.

Camilla: Uh- But with quantum technology also, you have to...

It's amazing, uh, opportunity for the world of resource optimization. So it's a lot of power. But as Peter Parker said, "With great power comes great responsibility." It was his uncle, though, but still.

Monica: Yeah. Um, okay, so coming back to agentic AI standards, so we have basically Visa, Mastercard, Google.

Everybody's creating [: 00:27:00

Camilla: I really hope that we're going to have an industry standard. Me too. Uh, th- this is a great opportunity for level the playing field also and not locking, uh, people into certain choices for the future. And hence, it would be great if we've seen it was great. It's a necessity that those standards are interoperable.

Uh, so we're sitting again at this cusp, as with so many other technologies, but I hope we learn from history that if we don't have standards, it costs the industry a great deal.

Monica: [: 00:28:00

Camilla: Mm.

Monica: So it's like when we have the standards, like interoperability within Southeast Asia when it comes to QR payments is amazing.

Camilla: Yeah.

Monica: Like, that's it, and it's ... That's a good use case, and the industry came together, so it's more of a, yeah, let's replicate those wins- Yeah ... in, in this segment.

Camilla: I think, uh, before we get to standards also, it's a lot of work to done in this space, so... And I know you're doing some of it, so that's exciting.

I'm going to watch to see where you're going. But it is, um, to get the taxonomy right so we all know what we're talking about, and defining the different step of payments to, um, initiation, authentication, settlement. All the different steps are actually well understood of how the payments move and where it goes from one actor to the next when the ball is passed on.

Monica: Yeah. Before we finish, what are you most excited about?

Camilla: I'm [: 00:29:00

So I, I do think that our industry, uh, can do so much good when we get it right.

Monica: Beautiful way to close because that's what I call the purpose-driven fintech stamp. Ta-da. Thank you, Camilla.