Ledger says AI is changing the economics of attacks

Ledger CTO Charles Guillemet said the biggest shift is not just that AI can help attackers, but that it is rapidly lowering the cost and effort needed to launch sophisticated attacks. In CoinDesk’s reporting, he said AI is making hacks cheaper and faster, which forces the industry to rethink how crypto systems stay secure. Ledger’s own recent security writing uses even stronger language, saying AI is collapsing attack costs and making older security assumptions obsolete.

That is an important point because crypto security has always depended partly on friction. If stealing funds or finding vulnerabilities takes enough time, skill, and money, many attacks never happen. Guillemet’s warning is that AI is eroding that barrier.

Hacks are not just getting bigger, they are getting easier to execute

The concern is not only about blockbuster exploits. It is also about the growing ease of vulnerability discovery, phishing, exploit chaining, and social engineering. Coverage of Guillemet’s remarks says AI is helping attackers identify weaknesses faster and package attacks more efficiently, which can shorten the path from finding an opening to stealing funds.

Ledger has been making a similar argument in its own publications. In February, Guillemet wrote that agentic AI creates new security problems because autonomous systems can propose and execute actions at scale unless the architecture clearly separates delegation from authorization. That becomes especially dangerous in crypto, where transactions are irreversible.

The warning lands as major exploits keep piling up

Guillemet’s comments are landing at a time when crypto security is back in the spotlight. CoinDesk linked the warning to recent protocol breaches including the massive Drift exploit and the Resolv collapse. Multiple summaries of the report also cited DefiLlama data showing that more than $1.4 billion in assets were stolen or lost in crypto attacks over the past year.

DefiLlama’s public hacks database also continues to track billions of dollars in historical losses across DeFi and bridge exploits, underlining how large the security problem already is even before AI’s impact is fully felt.

Ledger is pushing hardware and formal verification as part of the answer

Guillemet’s view is not that crypto is doomed, but that defenses need to get more serious. Summaries of his comments say he pointed to formal verification and hardware-based security as stronger defenses for wallets and protocols. Ledger’s own writing argues for hardware-rooted trust and for systems where humans still explicitly authorize sensitive actions instead of letting automation act unchecked.

That fits Ledger’s broader message over the past two months. Across its posts on AI risk, the company has been arguing that future-safe security will depend less on convenience-first software assumptions and more on architectures that preserve cryptographic control at the hardware layer.

Why this matters for the market

This matters because AI does not need to invent entirely new attack types to make crypto more dangerous. It only needs to make known attacks cheaper, more scalable, and easier to customize. If that keeps happening, more protocols, wallets, and users become vulnerable even when the underlying threat model looks familiar.

It also means crypto security can no longer be treated as just a smart contract auditing problem. The pressure now extends across wallet design, governance workflows, phishing resistance, operational security, and how humans interact with increasingly capable software agents. That is an inference from Guillemet’s comments together with Ledger’s recent AI security papers.

The bigger message is that security can no longer stay static

The clean takeaway is that Ledger’s CTO is warning about a structural shift, not a passing trend. AI is making it easier to attack crypto systems, and that means the old balance between attackers and defenders is getting worse. For the industry, the implication is simple: security models that were barely good enough before may no longer be good enough at all.