Coldcard Hack Bitcoin Explained: Are You Exposed?

The honest answer is: check, don't assume. You're potentially exposed if you own a Coldcard (Mk3, Mk4, Mk5, or the Q model) and generated your seed phrase — the 12-24 word backup that controls your bitcoin — any time between March 2021 and early August 2026. That's the window during which a bug in the device's firmware silently weakened how it created that seed. Updating to the patched firmware (5.6.0 for Mk4/Mk5, 1.5.0Q for the Q, 4.2.0 for Mk2/Mk3) stops the device from generating new weak seeds. It does nothing for a seed that already exists. If your coins live in an address generated before the patch, the underlying vulnerability travels with that seed regardless of what firmware version the device is now running.

No fifth wave of thefts has been confirmed since August 3, and roughly 90% of the stolen bitcoin is still sitting untouched in attacker-controlled wallets rather than being cashed out. That's arguably worse news than it sounds: it means the attackers who found this bug are patient, and it means researchers are still identifying which addresses were exposed. If you haven't checked your own device yet, you should treat this as unfinished business, not a closed case.

Hardware wallets like Coldcard are supposed to generate your seed phrase using a hardware random number generator — a physical chip designed to produce genuinely unpredictable numbers. That randomness is what makes a seed phrase practically impossible to guess; with true 128-bit entropy, there are more possible seeds than atoms in the observable universe.

The root cause traces back to March 2021, when Coinkite, the company behind Coldcard, integrated a software library called libngu and misconfigured a feature check in the process. Instead of always pulling from the hardware chip, some devices ended up routing seed generation through a deterministic software formula — one that produced a far smaller set of possible outcomes. On the Mk3, effective entropy collapsed to roughly 40 bits; on the Mk4, Mk5 and Q, it fell to around 72 bits. That sounds abstract, but in practice it meant an attacker with a laptop and no physical access to your device could work backward: generate the smaller universe of possible seeds, derive the bitcoin addresses each one would produce, and check those addresses against the blockchain for a balance. No theft of your device, no phishing email, no malware needed — just brute-force math against a bug that had sat undiscovered for over four years.

Once word of the flaw spread in late July, it wasn't one hacker who capitalized on it — at least 15 separate groups independently found and ran the same exploit, sweeping funds in waves that started 10 minutes apart, then 41 minutes apart, then over a full weekend, before a final wave on August 3 as more vulnerable addresses were identified.

Three steps, in order. First, check your device's current firmware version against Coinkite's advisory and confirm whether your specific model and seed-generation date fall inside the vulnerable window. Second, and this is the step people skip: even after updating firmware, treat any seed generated before the patch as permanently compromised. The fix protects future seeds, not past ones. Third, if you're affected, move your funds to a brand-new seed generated on the patched firmware — a fresh wallet, not a resettled one. Coinkite and the wider bitcoin security community have published migration guidance for exactly this; the point is that patching alone gives a false sense of resolution.

If you're not sure how your original seed was generated, or if the wallet has sat untouched since before August, the safe assumption is to migrate anyway. The cost of moving funds to a new seed is trivial. The cost of an already-guessable seed sitting exposed for months is not.

The base case going forward is a slow bleed rather than a dramatic sequel. No new wave has been confirmed since early August, and Coinkite's response — a patch shipped within days of disclosure, plus a public advisory — was about as fast as this kind of fix gets. That supports a bull case where most actively-used, at-risk wallets get migrated in the coming weeks and this becomes a contained, one-time hit.

The bear case is just as plausible, though. A meaningful share of vulnerable seeds likely sit in wallets their owners rarely check — inheritance setups, long-term cold storage, devices bought as gifts and forgotten. Those owners may never see the advisory at all, which means quiet losses could keep trickling in for months without making headlines the way the July wave did. And the roughly 90% of stolen bitcoin that hasn't moved is a genuine tail risk: if those attacker wallets ever start liquidating in size, it would reopen this as a news story and could add localized selling pressure to whatever market the coins land in.

The broader effect worth watching isn't a price move — this bug never should have, and didn't, meaningfully affect bitcoin's market price. It's a confidence question. Self-custody's entire pitch is that you don't have to trust a third party with your coins, but incidents like this show that pitch still depends on trusting the hardware and firmware you chose. Coverage at the time from outlets including CoinDesk flagged the risk that episodes like this nudge marginal, less technical holders back toward custodial products like ETFs, where a fund manager handles the keys.

That's a reasonable read, but it also somewhat misses the more useful takeaway for anyone who already holds a hardware wallet: check firmware, check your seed's origin date, and migrate if you're in the exposed window. The Coldcard hack is a reminder that self-custody security is not a one-time setup you complete and forget — it's a standard you have to keep verifying as vendors find and fix their own bugs.