Ledger CTO Charles Guillemet says AI is fundamentally shifting the economics of crypto security — and not in a good way.

"Finding vulnerabilities and exploiting them becomes really, really easy," Guillemet told CoinDesk. "The cost is going down to zero."

The warning comes in the wake of a damaging week for DeFi. Drift Protocol lost $285 million in a North Korean-attributed exploit, and yield protocol Resolv suffered $25 million in losses — part of over $1.4 billion stolen from crypto protocols over the past year, according to DefiLlama.

Why AI Changes the Math

Security has traditionally relied on asymmetry: attacks cost more than defenders' exposure. AI breaks that. Tasks that once took skilled researchers months — reverse engineering, exploit chaining — can now be done in seconds with the right prompts.

AI-generated code compounds the problem. As developers increasingly rely on AI tools, Guillemet warns that insecure code will propagate faster. "There is no 'make it secure' button," he said. "We are going to produce a lot of code that will be insecure by design."

What Helps

Guillemet advocates two approaches. First, formal verification — using mathematical proofs to validate code — which is more rigorous than traditional audits. Second, hardware-based isolation: devices like hardware wallets keep private keys offline and away from internet-connected systems.

He also flagged a growing threat: malware that scans compromised phones for wallet seed phrases, silently draining funds without user interaction.

His advice to protocol teams: "You need to be perfect." His advice to users: "You can't trust most of the systems that you use."