A significant security vulnerability in one of the most trusted names in self-custody hardware has surfaced, and the numbers attached to it are sobering. Coinkite, the manufacturer behind the Coldcard line of hardware wallets, has issued an urgent advisory instructing every user who generated a seed phrase on a Coldcard Mk3 running firmware versions 4.0.1 through 5.0.3 to treat their wallet as compromised and migrate all funds to a freshly generated wallet immediately. The warning did not emerge from routine internal audits — it followed a forensic investigation by Block's Bitcoin engineering and security teams, who traced up to 1,083 bitcoin drained remotely from wallets tied to the affected firmware range.
The scale of that figure — 1,083 BTC — deserves emphasis. At current market prices, that represents tens of millions of dollars in funds that moved without user authorization, from devices that many owners likely considered among the most secure ways to hold bitcoin. Hardware wallets occupy a privileged position in the self-custody ecosystem precisely because they keep private keys air-gapped from internet-connected environments. The discovery that Coldcard Mk3 units running specific firmware could produce predictable or weakened seeds strikes at the foundational promise of the product category itself.
What Block Found — and How
Block's security team did not stumble onto this by running abstract threat models. The discovery came through active investigation of real-world wallet drains — wallets that had been emptied remotely with no sign of physical compromise, phishing, or user error. That forensic starting point is critical context. It means the vulnerability was not theoretical; losses had already accumulated before the research surfaced the root cause.
The identified mechanism is weak seed-generation behavior. Seed generation is arguably the single most sensitive operation a hardware wallet performs. A properly implemented process draws on a high-quality entropy source to produce a 12- or 24-word recovery phrase that is, for all practical purposes, impossible to guess or reproduce. If that entropy is insufficient — if the randomness pool is shallow, biased, or predictable under certain conditions — the resulting seed may be reconstructable by an attacker with knowledge of the flaw. Block's team appears to have identified exactly that kind of weakness in the Mk3's implementation across the named firmware range, and the on-chain evidence of 1,083 BTC in drains suggests the vulnerability was not merely theoretical by the time it was documented.
The Scope of the Advisory
Coinkite's warning is deliberately broad. Rather than attempting to narrow the exposure to a specific subset of affected devices or usage patterns, the advisory covers every user who generated a seed on a Coldcard Mk3 within the firmware 4.0.1 to 5.0.3 window. That is a prudent posture: when the weakness is in the entropy generation process itself, it is impossible to tell after the fact which seeds were generated cleanly and which were not. The only safe assumption is that all seeds in scope are potentially exposed.
The practical implication is straightforward but non-trivial. Affected users must generate a new seed — on a different, verified device or a freshly updated and audited environment — and transfer all funds to wallet addresses derived from that new seed. Leaving any balance in a wallet whose seed was generated under the compromised conditions is to accept ongoing, unknowable risk. The attacker, or any party with access to the exploitation method, retains the theoretical ability to sweep remaining funds at any time.
A Stress Test for Hardware Wallet Trust
This incident is the most consequential hardware wallet security event in recent memory, and it raises questions that extend beyond Coinkite specifically. The self-custody hardware market has grown significantly as institutional and retail bitcoin holders alike have sought to reduce counterparty risk following a string of exchange failures. The implicit bargain in that shift is that moving custody to dedicated hardware eliminates the trust placed in third parties — but only if the hardware itself is trustworthy. A firmware-level entropy flaw rewrites that bargain in uncomfortable ways.
It is also worth examining how the vulnerability persisted across multiple firmware versions. The range cited — 4.0.1 through 5.0.3 — spans what appears to be a substantial deployment period. If weak seed generation was present across that entire range, the question of when the flaw was introduced and why it was not caught in internal testing or earlier third-party security reviews will require a transparent answer from Coinkite. The company's willingness to issue a public advisory and cooperate with Block's research is a point in its favor, but the hardware wallet community will rightly demand a detailed post-mortem on how this reached production firmware.
What This Means for Mk3 Holders
The immediate priority for anyone holding bitcoin on a Coldcard Mk3 is simple: check which firmware version was running at the time the seed was generated and, if there is any doubt, act as though the wallet is compromised. The 1,083 BTC already traced by Block represents confirmed losses, but the actual exposure among users who have not yet migrated is an open number. Every day a vulnerable wallet holds a balance is another day that balance is potentially at risk.
Beyond the immediate remediation, this episode reinforces a principle that gets stated often but is rarely tested this visibly: entropy quality is the bedrock of cryptographic security. Manufacturers, auditors, and the broader open-source community that scrutinizes hardware wallet firmware all have a stake in ensuring that future implementations are verified at the entropy level before they reach users. For now, the priority is migration — and speed matters.
Written by the editorial team — independent journalism powered by Bitcoin News.