A Coldcard firmware flaw tied to weak seed generation may have put thousands of Bitcoin wallets at risk, with investigators linking the bug to real theft and a growing pile of drained addresses.
- Weak seed randomness was the core problem
- TRM Labs says the flaw traces to Coldcard firmware 4.0.1
- Galaxy and TRM reported different loss estimates as the scope expanded
- Firmware updates do not fix old seeds
This is the kind of failure that makes self-custody believers grind their teeth. A hardware wallet is supposed to keep Bitcoin safe by keeping private keys offline. But if seed generation is broken, the whole security model can collapse. Fancy metal box, broken randomness, same old heartbreak.
According to TRM Labs, the issue goes back to a March 2021 Coldcard firmware version, 4.0.1. TRM says a bug caused some devices to generate wallet seeds with much weaker randomness than intended, cutting effective key strength from 128 bits to as little as 40 bits.
That number matters. Entropy is just a fancy word for randomness. In wallet security, more entropy means a much larger search space for an attacker. At 128 bits, guessing a seed is functionally out of reach. At 40 bits, the problem becomes dramatically smaller and, in some cases, attackable with serious resources.
So this was not a classic “remote hack” where someone popped a wallet over the internet. It was worse in a quieter way. A failure in the seed creation process may have left some wallets with predictably weak keys. If the seed can be reduced to a much smaller pool of possible values, an attacker doesn’t need to break the device. They just need to search the weakened space.
That’s the ugly part. A hardware wallet is only as strong as the randomness that feeds it.
The reporting on scale has shifted as investigators found more activity. The Hacker News, citing Galaxy Research, said analysts first observed a sweep of 1, 196 addresses in 41 minutes, followed by more waves. Galaxy later estimated 1, 367.05 BTC, worth about $88.6 million, across 4, 585 addresses. TRM Labs later put the drained amount at roughly 1, 816 BTC, or about $116 million, across more than 5, 200 addresses.
Those figures don’t line up perfectly, and that is not automatically a contradiction. It usually means the count changed as new wallets were identified, methods differed, and BTC pricing moved between reports. The important thing is the direction. The damage appears large, real, and still not fully pinned down.
One caution deserves to be stated early: the evidence supports a large-scale compromise tied to weak seed generation, but not every address in the various counts has been computationally confirmed as vulnerable. Galaxy said its on-chain analysis identifies the operator moving funds, not necessarily the theft itself. TRM was similarly careful, saying it had not computationally confirmed that every identified address was generated with weak Coldcard entropy.
That distinction matters. The funds were not “exposed” in some vague on-chain sense. The problem was much more specific. If the seed was weak, the private keys behind those wallets may have been vulnerable to brute-force recovery. That’s the part users need to understand.
TRM says attackers began draining wallets on July 30, 2026. According to its analysis, stolen funds have mostly been consolidated into a few attacker-controlled addresses, with limited laundering so far. TRM noted one 64.9 BTC deposit to Wasabi and 200 ETH sent to Tornado Cash on August 4, 2026, but otherwise saw mainly consolidation.
Wasabi is a Bitcoin privacy wallet often used to mix coins, while Tornado Cash is an Ethereum-based privacy tool that obscures transaction links. Both can be used for legitimate privacy, and both can also be used by thieves trying to muddy the trail. Here, the thieves appear to be sitting on a very visible pile of coins while they decide how to move it. Bold strategy. Not exactly subtle.
TRM also mentioned OP_RETURN spam messages sent to the hackers, including one offering to launder the funds for a 7% fee. For readers unfamiliar with the term, OP_RETURN is a way to embed small messages in Bitcoin transactions. So yes, even blockchain crime scenes get junk mail.
Another important point: updating firmware does not repair a seed that was already generated under the flaw. TRM says updates can help prevent new vulnerable seeds from being created, but they do not undo the damage for wallets already made with the bad randomness. Fox Business echoed that warning, saying affected users need to create a brand-new recovery phrase and move funds.
That is the practical takeaway, and it is brutally simple. If you generated a seed on an affected device and firmware combination, treat that seed as compromised. A patch is not a time machine. The old seed is the problem.
The bigger lesson here is not that self-custody is broken. It is that self-custody still depends on trust in the device firmware, the entropy source, and your own operational discipline. Hardware wallets reduce reliance on exchanges and cloud backups, which is a real win. But they are not magic. If one layer in the stack fails, the whole stack gets wobbly fast.
TRM’s analysis suggests the theft is still unfolding. It also notes that transaction patterns may point to multiple attackers, which means this may not be a neat, single-actor case. No authoritative source in the supplied reporting names the thief.
There’s a reason this kind of flaw lands so hard in Bitcoin circles. The whole point of self-custody is sovereignty: no bank, no exchange, no middleman. That remains the right instinct. But sovereignty without paranoia is just expensive optimism. If the seed is weak, all the cold storage branding in the world won’t save you.
Key questions and takeaways
-
What went wrong?
A Coldcard firmware issue reportedly weakened seed randomness on some wallets, making their private keys much easier to recover than they should have been. -
How serious is the damage?
Very serious. Reported estimates range from 1, 196 addresses in the first observed sweep to more than 5, 200 addresses in TRM’s later assessment. -
How much Bitcoin was involved?
Galaxy Research later estimated 1, 367.05 BTC, worth about $88.6 million, while TRM Labs put the figure at roughly 1, 816 BTC, or about $116 million. -
Can a firmware update fix an old seed?
No. An update may prevent new weak seeds, but any seed already generated under the flaw should be treated as compromised. -
Do we know who stole the Bitcoin?
No. The reporting does not identify a specific attacker, and TRM does not attribute the theft to a named actor. -
Are all the affected wallets confirmed?
Not entirely. Investigators have linked sweeps to wallets they believe were created with weak entropy, but they have not proven every single address in the counts.
If you used the affected firmware, stop trusting that seed and migrate now. In Bitcoin security, the hard truth is simple: once the randomness is gone, the wallet is already on borrowed time.
Further Reading
A few extra sources for the forensic side of this Coldcard mess:
