A Ledger executive says the recent Coldcard incident should not be read as evidence that hardware wallets are fundamentally unsafe. In his account, the central failure was a flaw in how wallet seeds were generated, not the use of dedicated hardware itself.
Ian Rogers, Ledger’s Chief Human Agency Officer, told Bloomberg that the breach stemmed from a 2021 firmware bug that routed seed generation through a software pseudorandom number generator instead of the device’s hardware chip. That reduced entropy to a level attackers could realistically search, turning what should have been an effectively unreachable key space into one that could be brute-forced.
What Rogers says went wrong
According to Rogers, the firmware issue caused some wallets to rely on software-based randomness rather than hardware-generated entropy. The result was a much smaller pool of possible seed values than users would expect from a hardware wallet designed to create highly unpredictable keys.
He said the affected setup left only 40 to 72 bits of entropy, a range small enough for attackers to search systematically. By contrast, Rogers said Ledger’s design uses a certified secure chip for entropy generation and has no software fallback, producing an address space he described as too large to brute-force.
The Coldcard theft timeline
The case drew attention after a major theft linked to compromised wallets. TRM Labs traced 1,082 BTC drained in the first 41-minute sweep on July 30, an amount cited in reports tied to the broader incident.
Rogers characterized the exploit as an example of how weak randomness can undermine otherwise strong security assumptions. In that framing, the problem was not that funds were stored on a hardware wallet, but that seed generation had been weakened by an earlier software flaw.
Why AI changes the threat model
Rogers said artificial intelligence is making this type of attack more dangerous by increasing the speed and scale at which attackers can search for weaknesses. He argued that AI does not create the original bug, but can dramatically improve an attacker’s ability to exploit limited entropy, exposed secrets, or coding mistakes.
He outlined three overlapping trends: AI gives attackers more firepower to identify vulnerabilities across many kinds of systems; AI-assisted software development speeds up code releases and can expand the attack surface; and companies are increasingly deploying agents that hold valuable internal secrets such as credentials, creating new points of failure if access is poorly controlled.
Broader concerns over agents and secret handling
Beyond crypto wallets, Rogers warned that the wider security challenge in the AI era is how much trust organizations place in software agents. Systems that are handed passwords, credentials, and other sensitive data may become highly attractive targets if they are not constrained by context and authorization.
He compared that level of access to handing car keys to a teenager: the issue is not only possession, but whether limits are built into how the system can act. Rogers said Ledger already offers tools that let an agent control a wallet without holding the underlying private keys, reflecting a security-by-design approach rather than relying only on policy.
What the incident points to next
The main lesson Rogers drew from the Coldcard case is that entropy quality remains a foundational security requirement. If randomness is weak, even strong-looking wallet designs can be undermined.
His broader argument is that AI will intensify pressure on every system built around code and secrets, not only crypto products. The next confirmed challenge for wallet makers and enterprises alike is to harden seed generation, reduce unnecessary secret exposure, and enforce tighter access controls as AI-enabled attacks evolve.
Source: beincrypto.com