AI lowers cost of crypto hacks, Ledger CTO warns

Artificial intelligence is significantly lowering the cost and difficulty of cyberattacks against cryptocurrency platforms, according to Charles Guillemet, chief technology officer at Ledger. In an interview with CoinDesk, Guillemet stated that AI tools can now perform tasks that previously required skilled researchers months to complete, in mere seconds, making vulnerability discovery and exploitation "really, really easy" and driving the cost to "zero."

This warning comes amid a surge in crypto heists. This week, the Solana-based DeFi protocol Drift was exploited, resulting in the loss of $285 million. The previous week, the yield protocol Resolv lost $25 million in a separate incident. Data from DefiLlama indicates that over $1.4 billion in crypto assets have been stolen or lost in the past year, a trend Guillemet expects AI to exacerbate.

The economic balance of cybersecurity is shifting, as AI erodes the traditional advantage where attacking a system was more costly than the potential reward. For crypto protocols, where code directly controls substantial funds, the need for absolute perfection is paramount. Guillemet noted that the increasing reliance on AI for code generation also introduces new vulnerabilities, stating, "There is no ‘make it secure’ button," and predicting the production of "a lot of code that will be insecure by design."

To counter these evolving threats, Guillemet advocates for stronger security measures beyond traditional audits. He highlighted formal verification, which uses mathematical proofs to confirm code integrity, and hardware wallets, which isolate private keys from internet-connected devices. He also warned everyday users to be skeptical of system security, advising, "You can’t trust most of the systems that you use," and anticipating a divergence between critical systems like wallets and protocols that invest heavily in security, and broader software platforms that may lag behind.