AI Summary
Fortune 的 AMA 汇集 Meridian X、Cortenair 等项目,讨论 AI 代理在链上金融中的安全、权限、托管与跨境依赖问题。
- 活动由 AI 预测衍生品平台 Fortune 主持,主题聚焦 AI 代理对链上金融安全的影响:是让金融更易保护,还是因自主性增强而更难防御。
- Meridian X 的 Harry 介绍其基于稳定币的中间清算层,利用区块链结算提升全球航空票务效率,连接上游航司与下游 OTA。
- Cortenair 的 Vitri 介绍其 AI 原生分布式云基础设施,目标是让 AI 计算更高效、可扩展、可靠且可验证。
- 主办方强调产品责任应让普通用户无需安全工程师背景,也能理解自己批准了什么、如何更改、出问题找谁。
- Fortune 将新闻、监控、研究、情绪与概率分析、订单簿分析等 AI 工具整合进预测市场体验,并用多模型交叉验证观点。
- 讨论计划从托管与人员入手,再拆解交易所背后的流程,并探讨 AI 代理竞赛、跨境依赖及市场信号能否作为早期预警。
- 主持人指出 2016 年 CEX 黑客攻击更多针对托管基础设施、私钥管理和内部访问控制,而非智能合约代码,并以此引出对跨链桥模式的反思。
建议用户在选择任何 AI 金融产品前,先明确其权限范围、托管方式和应急支持渠道。
Transcript
angles. We'll start with the custody and the people and then remove the processes behind
the exchange and we'll explore the race around AI agents, the dependency that across
borders and whether the market signals can contribute to an early warning. And finally,
we'll look at the technologies before I rely on and the evidence we should ask for.
So as a founder, I'm also interested in making those responsibilities more understandable.
A user does not need to be as security engineer to ask what they have approved, how they
can change it and who can help when something goes wrong. And I hope we can live people
with practical questions when they feel comfortable bringing to any product they use.
And for anyone joining Fortune for the first time, we're an AI power prediction
derivatives hub. We bring prediction market liquidity to get her through one interface.
So people can explore market without having to move between separate platforms at every
step.
For some agent brings AI tool into that experience with news, monitoring, research, sentiment,
and probability analysis and also order book analysis. So we have multiple models to cross-check
their views as part of that process. And our interest today topic comes from the responsibility
of bringing information closer to a financial decision. And for me, I think an explanation
should have somewhat understand that today are using and the choice in front of them.
That is why I'm looking forward to hearing how other themes approach the permissions, security
and experience around them. There's useful experience on learning here on how people
put building products and for the communities that are using them.
So we have several projects joining us back today with their different perspectives
across the ecosystem. Let's meet everyone before we get started. So please start with a little
part of yourself, your role and the work of your project does. Then tell us how your experience
connect with today's topic. And also one short heart attack will be great. Do AI agents
make on chain finance easier to secure or just giving them more autonomy make the whole
system harder to defend?
First we have from Meridian X. It's lovely to have you back with us. Please introduce
yourself. That's about your role and your work. Please go ahead.
Good morning everyone. Greetings from Canada. I'm Harry and I'm the core contributor from Meridian
X. So let me introduce us briefly. So Meridian X is a middle claiming layer based on the
stablecoins. And we utilize the blockchain settlement system to increase the efficiency of
the global aviation industry. So in similar terms, we buy wholesale and sell retail airplane
tickets from connecting upstream airlines to downstream online travel agencies such as
strip.com bookings.com, a quarter, etc. And in a very simple term, we just we are just trying
to make your airplane tickets cheaper by utilizing about three technologies. And I'm very excited
for today's AMA and thanks for your global point inviting me.
Okay. Thank you Harry. And now let's move to victory. Hello, please introduce yourself,
your work and your housing. Hello everyone. My name is vitri. I'm excited to be here
representing Cortenair. In Cortenair, we're building an AI native distributed cloud in
infrastructure designed to make computing more efficient, scalable, reliable and verifiable.
Now our goal is to make AI computing more accessible, cost efficient and trustworthy for developers
businesses and the next generation of autonomous AI application. So I'm happy to be part of this
discussion because I believe the future of AI depends not only on how intelligent our system
becomes but also on how securely and efficiently they operate. So thank you for having me.
Fuvati, you host. Okay. Thank you, victory. So while we're waiting for other speakers to join us,
I think we can move to the round table question. Oh, there's Cort card joining us.
I think you can introduce yourself, your word and your housing for today's topic.
Okay, we got silence. Cort card, can you hear me? Okay, let's move to the round table detail. So
to the round table discussion, we are using the raised hand method. So you can raise your hand
while you like to answer or add another few. And then I will bring you one at a time and invite
people directly as well. So there's a lot of room for different opinions. So
where we discuss and incident or defense, we can you can give some examples or what you've seen in
practice. So we can help everyone to understand what we're talking. So let's start with the system that
people already trust to look after their assets. Before we talk about an autonomous agent,
I would like to understand what exchange bridges can teach us about the access, the custody,
and the processes around them. So the CEX hacks in 2020 six has targeted custody infrastructure,
private key management, and internal access control, more than smart contract code. What does the
pattern of these bridges tell us about where the industry real security depth shifts and why
happened to the lesson from previous cycles actually implemented at scale. So we would like to get
started on this one. Okay, Harry, go ahead. Yes, the CEX hack has spent a
like a huge issue for our Web3 communities and exchanges. I believe the first hack happened back in
2013, like until a few days ago, over 15 million US dollars has been hacked. And like there's
various reasons and the attacks are from many different organizations or different people.
And like, I believe this, I mean, the recent Binance hack was from, I mean, some people say it's
from North Korea, because since they have a strong hacker team name like Lazarus, I believe.
And but by the way, like there is something for us to consider, I mean, to ponder since traditional
brokers such as like interactive brokers, fidelity, etc. Like these traditional brokerages who
allows us to buy and sell stocks stocks were very rarely hacked. So I feel like our centralized exchanges
needs to like try to do something the traditional exchanges are doing in order to improve their
ability to not get hacked. And maybe have more spending on security, I think, or do more
KYC or like methods like this. And because like every time an exchange get hacked, the
how to say that, like, I mean, about three participants' confidence about the technology is
going to be hit super hard. So like, I feel like if these hacks keep happening, then
it's going to become a huge issue for our community to grow fast. And that's my take back to you.
Thank you. Yeah, thank you. I agree with your mentioning the finance hack. And we can take a look
at the how they respond after a problem is detected and how the team recognized the
concert and decided what to do. We would want to know who can intervene and which actions are
available and how the team avoids creating another problem while responding. And I think we also need
to be aware to communicate what is known without making promises that the investigation cannot yet
support. And I think from the user's perspective, the process can matter as much as the initial
alert where people need to understand whether their access or provisions are affected. And
where to find reliable information. So yeah, I appreciate bringing those responsibilities into
conversation because it shows how it seems handles a typical situation once the ordinary process
has been disrupted. So I saw victory raising her chance. Do you want to act anything?
I really appreciate what the first speaker has said. And I think one of the biggest
lessons from exchange security breaches is that blockchain security goes beyond smart contracts.
For a long time, the industry focused heavily on auditing smart contracts because they are visible
and directly connected to transaction. However, the security of the infrastructure behind those
transactions is equally important. So we consider the private key management employee access
permissions, cloud infrastructure and internal system. Even if a smart contract is perfectly
edited weaknesses in these areas can still expose users from. So that is my take on it.
Over to you, hosts. Okay, thank you for adding in victory. Yeah, I agree that when we
compare the responsibilities that enforce, I think it is useful to look at
a world control and action and with dependencies that remain rather than deciding
the race from a category level alone. And while we are evaluating an exchange or wallet or a
product call, we may be relying on different arrangements and the thing is we should understand
those arrangements on their own terms like which permissions access and what can be changed and
also what happens when a company fails. So that gives us a more practical comparison that
assuming on one architecture that removes every problem. So we should know with risk that
design changes and with responsibility to move somewhere else. Okay, thank you. So we also have
I see be joining us today. You can introduce yourself, your word and the logic for today's topic.
And you can. Hello, everyone. Yes. Okay, okay. Hello, everyone. Can you guys hear me? Yes,
we can hear you clearly. Okay, okay. Okay, okay. So hello, everyone. I'm Azizbi. I'm the CM of
Sonia. It's actually a great opportunity to be here to join this conversation and I'm really,
really sorry for being late. I was having some network issues. So yeah, Sonia is basically built
around a simple idea, which is, which is as the AI makes digital content easier to like create
we also need better ways to establish ownership, protect its identity and give it real value.
So from our side, what we are building is what we call the AI Music Asset Ephonomy.
So we are connecting AI Music creation and with with assetization, digital ownership and provenance
discovery, collecting, trading and licensing. So we are also seeing AI agents playing an important
role in the ecosystem, helping people discover music evaluate assets, manage collections and connect
with creators with new opportunities. So yeah, that was all about me and Sonia. What are you host?
Okay, thank you Azizbi. Do you have any take from our first?