Artificial intelligence is dramatically tilting the economics of crypto security in favor of attackers, Ledger’s CTO warns — and the timing could not be worse.
Charles Guillemet, chief technology officer at hardware wallet maker Ledger, told CoinDesk that AI tools are making it faster and cheaper to find and exploit vulnerabilities, eroding the long-standing security assumption that hacking should be harder and more expensive than defending. “Finding vulnerabilities and exploiting them becomes really, really easy,” he said. “The cost is going down to zero.”
Why it matters now
- High-profile exploits are mounting. This week Solana-based DeFi protocol Drift lost about $285 million in an exploit — one of 2026’s largest so far — and a week earlier yield protocol Resolv was hit for roughly $25 million. Over the last year, DefiLlama reports more than $1.4 billion in crypto was stolen or lost to attacks.
- AI accelerates traditionally slow, technical tasks. Activities that once took expert researchers months — reverse engineering binaries, linking multiple vulnerabilities into a single exploit chain — can now be accomplished in minutes or seconds with the right prompts.
- AI-generated code amplifies risk. As developers lean on generative tools, insecure patterns can propagate rapidly. “There is no ‘make it secure’ button,” Guillemet said. “We are going to produce a lot of code that will be insecure by design.”
What this means for protocols and users
- The stakes are higher for blockchain systems that control significant pools of funds. “You need to be perfect,” Guillemet warned teams building protocols — because small bugs in smart contracts or wallets can be catastrophic.
- Traditional audits may no longer be sufficient. Guillemet recommends stronger defenses such as formal verification — using mathematical proofs to validate code correctness — which can catch issues audits miss.
- Hardware-based security becomes more important. Devices that isolate private keys from internet-connected systems (hardware wallets) reduce exposure, he said: “When you have a dedicated device not exposed to the internet, it is more secure by design.”
- Malware is getting smarter. Guillemet described attacks that scan compromised phones for seed phrases and drain funds without any user interaction, underscoring that compromise can start outside the blockchain layer.
Practical takeaways for users and teams
- Assume systems will fail: “You can’t trust most of the systems that you use,” Guillemet said. Treat wallets and services accordingly.
- Move high-value holdings to cold storage and use hardware wallets for key custody.
- Harden operational security: minimize where seed phrases and private keys are stored, enable multi-factor and multi-signature protections, and segregate sensitive workflows from general-purpose devices.
- For teams: invest in formal methods, continuous security testing, and threat modeling that anticipates AI-augmented attackers.
A widening security divide
Guillemet expects a split: mission-critical systems such as wallets and major protocols will invest heavily and adapt, but much of the broader software ecosystem may struggle to keep pace. “It’s really easier to hack everything,” he said.
Bottom line: AI is lowering the barrier to attack across the crypto stack, forcing developers, platforms and users to rethink defenses — from verified smart contract design to hardware isolation and stricter operational hygiene — before the next major heist.
Read more AI-generated news on: undefined/news
