CZ: AI Crypto Hacking Fears Overstated, Not Direct Threat

Everyone worried about AI hacking crypto, but the internet and banking systems rely on weaker encryption, Zhao wrote. That said, AI doesn’t need to break Bitcoin’s cryptography to cause damage; leaked data already enables more phishing, social engineering, etc.

His point rests on a split between two very different things: AI breaking cryptographic systems, and AI making existing cybercrime work better. Bitcoin’s security relies on robust cryptographic mechanisms, and AI getting better does not make them inherently vulnerable. What it does change is the quality of old tricks, from fraudulent emails and fake investment pitches to convincing impersonations.

Zhao named phishing, social engineering, and the misuse of leaked data as the more immediate threats. Human factors, he said, remain the main weak point for crypto users, even as AI advances.

AI-Linked Scams Earn 4.5X More Per Operation

Industry data points the same way. Blockchain analytics firm Chainalysis flagged the growing use of AI tools in its 2026 Crypto Crime Report, published January 13, 2026. It found that AI-enabled crypto scams generated 4.5 times more revenue per operation than scams without such links. Here is how the two groups compare:

• Average revenue per operation: $3.2 million for AI-linked scams, against $719,000 for those without identifiable AI links
• Median daily revenue: $4,838, against $518
• Transfers per day: 35.1, against 3.89

Deepfakes, face-swapping software, and large language models make these schemes more convincing. Overall, crypto losses from these scams are projected to reach $17 billion in 2025.

As AI tools become easier to get, the contest between criminals and security providers will intensify. Scammers can automate communications and build fake identities with less effort, which makes it harder for investors to tell legitimate messages from fraud. AI cuts the other way too. Exchanges and analytics firms can use it to flag suspicious transactions and fraudulent wallet addresses, and to trace stolen assets faster. How well crypto security holds up will depend on both blockchain resilience and the industry’s ability to counter AI-enhanced fraud.

Hashlytics Take

CZ is right about the cryptography, and it barely matters to the person who loses money. Funds almost never leave a wallet because someone broke elliptic curve math. They leave because a human approved a malicious transaction or handed over a seed phrase, and AI is very good at manufacturing the moment that makes someone do that. The Chainalysis numbers are the stronger evidence here, more than the post itself. One caveat is worth keeping in mind: the comparison is against scams “without identifiable AI links,” so the AI group is only the operations where researchers could spot the tooling. Those are likely the bigger, more organized outfits, which may flatter the 4.5x gap. The direction is hard to argue with. The exact multiple is softer than it looks.