Ledger says a recent exploit affecting Coldcard is a warning that hardware wallet security depends heavily on how wallet seeds are generated. According to Ledger CTO Charles Guillemet, the incident exposed how a failure in randomness can undermine the protections users expect from a dedicated signing device.
The company describes the flaw as a case where Coldcard firmware fell back to software-based randomness instead of relying on the device’s hardware random number generator when creating recovery seeds. Ledger says that made some private keys guessable and enabled theft, turning a technical design choice into a direct security risk.
Why Ledger says the flaw mattered
Guillemet presents the Coldcard incident as more than a product-specific bug. In Ledger’s view, it highlights a core assumption behind hardware wallets: that the randomness used to create a seed must be strong enough that no attacker can predict it.
If that process is weakened, the security model can break down even if the wallet is otherwise designed for offline key storage. Ledger says the exploit showed how a software fallback can create that weakness, because the resulting entropy may be more predictable than users realize.
Ledger contrasts its own design
Ledger says its devices were not affected. The company states that its wallets derive entropy from a certified hardware random number generator housed inside a Secure Element, and that there is no software fallback in the seed-generation path.
According to Ledger, every seed is produced with the full 256 bits of entropy. The company argues that wallet randomness should come from a physical source and be independently certified, rather than relying only on vendor claims about how a system behaves under edge cases.
AI is changing the threat model
A central part of Ledger’s argument is that artificial intelligence is speeding up the discovery of software weaknesses. The company says attackers can now use AI systems to scan code and identify vulnerable patterns at machine speed, which increases the pressure on wallet makers to remove avoidable design risks before products reach users.
In that framing, defenses also need to become more rigorous. Guillemet argues for security-by-design anchored in hardware protections and mathematical guarantees, rather than assumptions that vulnerabilities will be caught later through normal review processes.
Certification versus review
Ledger also draws a distinction between open-source development, peer review, and independent certification. It says these are not interchangeable, especially for components as sensitive as random number generation.
The company’s position is that transparency and outside review remain useful, but they do not by themselves prove that a wallet’s entropy source is trustworthy in all conditions. For that reason, Ledger argues that independent certification should play a larger role as AI reshapes both offensive and defensive security work.
What users can verify next
Ledger says one practical lesson from the exploit is that users should look beyond broad security claims and understand how a wallet generates randomness for recovery seeds. In particular, the company points to whether entropy comes from hardware and whether that process has been independently certified.
The broader next step, as Ledger frames it, is for wallet security models to adapt to an environment where AI can accelerate vulnerability research on both sides. In that setting, the way randomness is produced and verified becomes a more important part of judging whether a device is resilient.
Source: decrypt.co