Bank cards, ATMs, Tap to pay… is AI the next evolution in banking?

Welcome to AI Collision 💥,

In today’s collision between AI and our world:

• 15 years of bitcoin…now with AI
• Stripe AI and fraud-free banking
• Saylor on new AI made securities

If that’s enough to get the AI transacting, read on…

AI Collision 💥

When I first started researching bitcoin in 2010 it took me a little while to fully appreciate that it would completely change the global financial system. But it was relatively clear that a decentralised, distributed alternative monetary system was a powerful idea and one that for generations to come would provide an opt-out from a legacy system that was fundamentally flawed and rigged against the average person trying to make their way in the world.

I still hold the belief that crypto, starting with bitcoin is going to change everything. But I always took the view that it would take generations, and that beyond our lifetime to really see a fully-ledged bitcoin reserve global financial system.

But with AI, and the speed at which AI is developing, I think that we’re likely going to see an acceleration of the adoption, and integration of crypto and bitcoin in a global financial system, both in the legacy (TradFi) system and the decentralised (DeFi) system.

What I hadn’t banked on back in the 2010s was how much AI would change banking, finance, and ultimately crypto and bitcoin.

But in the decade or so of research I’ve put into AI, it’s been clear that it too would change the very fabric of society, from the outset. It was obvious that as AI got better as would its ability to perform the tasks of humans, particularly white-collar jobs, things like lawyers, accountants, writers (eek!), researchers, and a whole host of other “creative” based jobs.

It will disrupt jobs, make certain roles redundant, but also enable a gigantic productivity leap for others who can augment their work with AI to be better, faster, smarter, more efficient, productive and accelerating global growth along the way.

And we’re starting to see what that looks like right now. There are a slowly increasing number of examples of AI in use in areas not first considered by the wider market that is fast changing how the entire global financial system works.

Some people might say that’s going to create a systemic risk to the global financial system. Regulators will no doubt want to step in to quash the use of AI in the system. But it seems like AI is doing far more good work now in banking and finance than harm. And quite possibly if AI is used in the right way, it might even make the position of regulator redundant and prove to be the ultimate in consumer protections.

The following is full text of a tweet from Gautam Kedia at Stripe. He’s the lead AI and Machine Learning executive. He says,

TL;DR: We built a transformer-based payments foundation model. It works.

For years, Stripe has been using machine learning models trained on discrete features (BIN, zip, payment method, etc.) to improve our products for users. And these feature-by-feature efforts have worked well: +15% conversion, -30% fraud.

But these models have limitations. We have to select (and therefore constrain) the features considered by the model. And each model requires task-specific training: for authorization, for fraud, for disputes, and so on. Given the learning power of generalized transformer architectures, we wondered whether an LLM-style approach could work here. It wasn’t obvious that it would—payments is like language in some ways (structural patterns similar to syntax and semantics, temporally sequential) and extremely unlike language in others (fewer distinct ‘tokens’, contextual sparsity, fewer organizing principles akin to grammatical rules).

So we built a payments foundation model—a self-supervised network that learns dense, general-purpose vectors for every transaction, much like a language model embeds words. Trained on tens of billions of transactions, it distills each charge’s key signals into a single, versatile embedding. You can think of the result as a vast distribution of payments in a high-dimensional vector space.

The location of each embedding captures rich data, including how different elements relate to each other. Payments that share similarities naturally cluster together: transactions from the same card issuer are positioned closer together, those from the same bank even closer, and those sharing the same email address are nearly identical.

These rich embeddings make it significantly easier to spot nuanced, adversarial patterns of transactions; and to build more accurate classifiers based on both the features of an individual payment and its relationship to other payments in the sequence.

Take card-testing. Over the past couple of years traditional ML approaches (engineering new features, labeling emerging attack patterns, rapidly retraining our models) have reduced card testing for users on Stripe by 80%. But the most sophisticated card testers hide novel attack patterns in the volumes of the largest companies, so they’re hard to spot with these methods.

We built a classifier that ingests sequences of embeddings from the foundation model, and predicts if the traffic slice is under an attack. It leverages transformer architecture to detect subtle patterns across transaction sequences. And it does this all in real time so we can block attacks before they hit businesses.

This approach improved our detection rate for card-testing attacks on large users from 59% to 97% overnight.

This has an instant impact for our large users. But the real power of the foundation model is that these same embeddings can be applied across other tasks, like disputes or authorizations.

Perhaps even more fundamentally, it suggests that payments have semantic meaning. Just like words in a sentence, transactions possess complex sequential dependencies and latent feature interactions that simply can’t be captured by manual feature engineering.

Turns out attention was all payments needed!

This part in particular shocked me, “This approach improved our detection rate for card-testing attacks on large users from 59% to 97% overnight.”

If you’re unaware, Stripe is a financial technology company, founded by Patrick and John Collison that has become one of the world’s most valuable financial companies. They provide the payment “rails” so to speak for millions of businesses and giants like Amazon and Shopify, enabling anyone big or small to integrate easy payment technology into their businesses.

Stripe is making headlines already as it is, with an announcement they’re going to start offering stablecoin accounts, which is a significant leap forward in the acceptance of crypto networks into an already massive global financial giant.

If you can roll up and roll out advanced AI to monitor and protect payments from fraud with the success rate that Kedia talks about, then very quickly you can eliminate one of the main criticisms of crypto from sceptics…that it’s too risky and full of fraud for the average consumer.

Combining AI with decentralised finance, all of a sudden looks like an incredibly gigantic opportunity.

And then there’s the way in which AI has been used to create entire new kinds of securities. Take a look at this from Michael Saylor:

He’s talking about AI created securities. The way in which he talks about his use of AI though is EXACTLY what I’ve been talking about. It’s clearly not replacing him, but it’s extracting the maximum creative capacity from what he’s trying to do with Strategy.

And that’s how AI changes the game in global banking and finance. It doesn’t kill off all employees from the banking system, but it enables possibility around protections and innovation that we perhaps never really thought possible.

And when you start to layer that on top of the idea of whole new capital market restructuring that bitcoin and crypto is enabling, the acceleration of that I think will take a lot of people by surprise.

Even Max Keiser, one of the most bitcoin-maximalist people in the world, has seen just what this might lead to:

What does it mean for you? Well, I think a company like Stripe (currently private) will eventually IPO and list on the market. And I think it makes for a very enticing investment proposition if it does.

And then when I look to companies like Strategy and what they’re doing now in the space with bitcoin and the use of AI in their company, it’s hard not to think they’re leaning into this new technology to maximise the potential of their company for shareholders.

Ai and banking and finance and bitcoin and crypto…there’s a lot going on there, but it looks like it’s about to unlock what could amount to trillions in capital and a huge investment opportunity.

#Ad

Biden tried to chip it away… but ended up carving Trump.

America’s Hidden $150 Trillion Fortune is being revealed.

And you don’t need to be American for the chance to benefit.

Capital at risk

Boomers & Busters 💰

AI and AI-related stocks moving and shaking up the markets this week. (All performance data below over the rolling week).

Boom 📈

• Vertiv (NYSE:VRT) up 11%
• Everspin Technologies (NASDAQ:MRAM) up 10%
• Microsoft (NASDAQ:MSFT) up 10%

Bust 📉

• Apple (NASDAQ:AAPL) down 7%
• Palantir (NASDAQ:PLTR) down 6%
• Alphabet (NASDAQ:GOOG) down 5%

From the hive mind 🧠

• I’ll be first to admit, I don’t use search like I used to anymore. The way in which I tend to look for information now is initially via ChatGPT or Grok, or Gemini. This is going to change search, Google’s dominance, and how Apple, Microsoft and every other big tech company generates revenues online. AI is very much changing search, and it’s hard to see who comes out on top (for now).
• I’m all for the use of AI to bring back memories, like animating old photos and things like that. But this… nah. This is just weird.
• I believe AI will be revolutionary in aged care. Combined with robotics of course, but I think it will help people live longer, safer, and a better quality of life in older age.

Artificial Polltelligence 🗳️

Weirdest AI image of the day

While they choose a new Pope, let’s look back on past presidential visits to the U.S.

ChatGPT’s random quote of the day

Thanks for reading, and don’t forget to leave comments and questions below,

Sam Volkering

Editor-in-Chief

AI Collision