The Fat Finger Syndrome

I was intrigued to read that a major cryptocurrency exchange in South Korea, Bithumb, had an internal error where an employee sent customers a prize that should have been 620,000 Korean won (around $423) to 695 qualifying customers, but instead credited accounts with 620,000 bitcoins – equivalent to over $40 billion.

How a major crypto exchange gave away $40bn

Whoops!

The good news is that the exchange managed to capture 99.7% of the erroneous credits, but it still lost millions.

It made me reflect on what we call the fat finger syndrome in the financial industry. Is this the worst example ever committed?

No. There are many. In fact, another South Korean fat finger occurred in 2018 with Samsung, when an employee mistakenly gave staff 1,000 shares each instead of 1,000 won (less than $1) per share. The total cost was a notional value of over $105 billion. Some employees then compounded the error by selling the non-existent stock, causing the company's share price to plummet.

But perhaps the worst example was with Citibank in 2022 when one of their traders tried to sell a basket of equities worth $58 million but accidentally entered the quantity in the wrong field, creating an order for approximately $444 billion. It cost the bank $48 million to sort out the error and a lot of internal wrangling.

What this demonstrates is that the biggest exposure in financial services today is the humans, not the systems. It is clear that human error is the biggest crack in the system. We see this regularly with authorised push payments and romance scams. It is social engineering and dumb people who mess up the systems.

The friction here, however, is that the systems are here to support the humans, not the other way round. So, why are systems processing strange and suspicious transactions that humans enter without asking: why?

This is probably going to be the biggest change in the next few years, as AI will spot the wrong entries when someone says bitcoin rather than won or a payment to an address that is suspicious. In fact, the biggest developments in AI will be AI to scam and AI to identify scams. This will change the way we live a lot.

It is part of what I call The Intelligent Bank.

The Intelligent Bank will know when things are going wrong, processed incorrectly, looks like a human mistake and more. It will challenge us to confirm that this is what we really want to do but, unlike today’s systems of accepting cookies and t’s and c’s, it will point out why it’s asking that question.

This is what annoys me about GDPR and more. You get asked if you accept the cookies or terms and conditions and you just click yes, as you cannot be bothered reading such things. What about an AI that said if you click now, this is what it means to your privacy and data or if you click now, you are contractually bound to an 18-month contract. You get the idea.

What we really need in life and in finance is far more intelligent systems that talk to us like humans rather than machines, and process transactions with human insights and not just transaction processing.

This is part of a discussion I’m having with Xpand IT on Wednesday, 25th February 2026. It’s going to be a free webinar starting at 15:00 GMT, and you can register here https://lnkd.in/e6zSHA8S (for free!).

See you there.