Quick Take
  • On October 29, attackers took about $6.3 billion from Binance, the largest theft in crypto history and roughly four times the previous record.
  • Their way in began months earlier, with an engineer searching for an AI coding tool.
  • In the days after, something happened that I had not seen since Bitcoin hit its all-time high in October 2025.
  • People who have nothing to do with crypto started asking me what was going on in it.

What Happened

Nobody has shown that AI planned or ran this attack, and I will not claim it did. But AI was the bait at the start, and a tool in the middle, and attackers are adopting it quickly. TRM Labs counted a record 201 crypto hacks in the first half of 2026, against 83 a year earlier.

Investigators at Elliptic later found the thieves creating their own meme tokens to wash part of the money. It was a small share of the total, but it is how this industry looks from outside right now, with meme coins on one side and exploits on the other.

Market Context

I think insiders have not fully taken that in. We measure an event by what it does to prices and withdrawals.

So it is a fair moment for a basic question. AI is changing how software gets written, how markets get traded, and how attacks get carried out. What is crypto for in that world? The answer starts with bad news and gets better.

That is the long-term case. The practical one, for anyone with an exchange account today, is what AI does to their trading.

Prediction markets give a second data point, this time with real users. A Bloomberg analysis of about two million Polymarket wallets found that retail traders picked the correct outcome more often than the bots did and still lost, because they entered late at worse prices.

Why It Matters

On October 29, attackers took about $6.3 billion from Binance, the largest theft in crypto history and roughly four times the previous record. Their way in began months earlier, with an engineer searching for an AI coding tool.

In the days after, something happened that I had not seen since Bitcoin hit its all-time high in October 2025. People who have nothing to do with crypto started asking me what was going on in it.

Details

A better measure is whether it leaves our corner of the internet, and this one did. For a lot of people, it is the first thing they have heard about crypto in a long time.

So Far, AI Has Mostly Been Used Against Us

Binance has published how it happened. In June, the wallet engineer searching for that AI tool installed a tampered package instead.

In September, another engineer went through fake job interviews on live video and was handed a coding assignment carrying malware that no security product had seen before.

From his machine, the attackers used one of Binance’s internal AI tools to reach two more engineers and then the wallet-signing service. The FBI has attributed the theft to North Korea.

We have been on the receiving end ourselves. Phemex was breached in January 2025, and we rebuilt our security architecture because of it.

Binance says it has replaced nearly all of the affected assets from its own funds, and an independent review is still to come.

If that is all crypto is in a few years, there will not be much of an industry left. I do not think that is where this goes, and AI is a large part of why.

The One Thing AI Cannot Make More Of

AI makes digital things abundant. Code, apps, images, music, and soon whole films can be produced by anyone in minutes, and when something becomes that easy to make, its value heads toward zero.

Bitcoin is the opposite kind of digital object. Ask a model for a thousand logos, and you have them in a minute. Ask it for one bitcoin, and it has to buy it like everyone else. No model can produce another coin or rewrite the ledger that records who owns it.

The same applies beyond Bitcoin. As generated pictures and video fill the internet, people will want digital goods that are scarce enough to hold value, and a record of a photo or a video that cannot be doctored afterwards.

NFTs and other blockchains can provide both. Much of the NFT crowd has since moved on to meme coins or physical collectibles, which answer the scarcity question in different ways, but the question itself is not going away.

AI agents are part of this too. As I argued here in February, they will need money they can hold and move without asking a bank, and crypto already built it. Scarce assets, records nobody can edit and open rails are what this industry has that AI cannot generate.

Where AI Helps a Trader

A year ago, a research lab called Nof1 gave six leading AI models $10,000 each to trade crypto perpetuals live with no human help. Four of the six lost money, and the worst lost about 60% in a little over two weeks.