So just last week, someone emptied almost 1,200 Bitcoin addresses in about 41 minutes. The crazy part is that they did not phish anyone, did not compromise anyone’s computer, and never touched a single one of the affected devices. They did not even need to know who the victims were. All that happened was that attackers worked out what the private keys were behind those wallets.
The kicker: all of those private keys had been created on Coldcard hardware wallets. That became possible because of a bug that had been sitting in its firmware since March 2021. The randomness meant to generate those private keys was nowhere near as random as it should have been. These were low-entropy seeds.
If you are one of the people who lost funds this week, I’m really sorry for your loss. There is a line from Ryan Castellucci, who did a lot of early work on brainwallet weaknesses, that still holds true: just because someone’s setup is bad does not give anyone the right to take their money. Do not blame the victim. This is not the time to be a jerk about it.

What actually happened with Coldcard
The short version is that a firmware random number generator weakness produced low-entropy seeds. This was not phishing. It was not a stolen device. It was not someone physically tampering with your hardware wallet. The problem was in the quality of the randomness used to create the seed material.
That matters because entropy is what keeps your wallet unique. When people set up a new wallet, I often get asked: “How do I know my seed phrase or private keys are unique? Does my wallet somehow check that this seed has not been used before?” The simple answer is no. Your wallet does not meaningfully check whether your exact seed has appeared somewhere else in the past. What keeps us safe is entropy, and having a lot of it.
Entropy is just a technical way of talking about how much randomness went into making your seed, measured in bits. If everything works correctly, that randomness is what makes brute-forcing your keys practically impossible. Or at least that is the promise. As we saw this week, that promise is not always delivered by wallet vendors.

Why weak seeds are so hard to detect after the fact
This is what makes entropy failures especially nasty compared with most other wallet problems. A weak seed and a strong seed can look exactly the same. You cannot just stare at twelve or twenty-four words and tell whether they were generated from 128 bits of good randomness, one bit of bad randomness, or some broken fallback path inside firmware.
The only realistic way to know that a seed coming out of a wallet is any good is by understanding the input entropy that went into making it. That is why I have long preferred wallets that mix multiple entropy sources together. If one source turns out to be weak, you still have another source mixed in. You get defense in depth at the exact layer where failure is unrecoverable.

This also helps explain why some people reported doing everything “right” with their Coldcard and still losing funds. If the firmware was generating low-entropy seeds, then two different users could end up with the same seed purely by random chance. From the outside, it can look like bad luck or user error. But if the randomness source is broken, that kind of collision becomes much more likely than it should be.
Air-gapping does not fix entropy quality
One trend in hardware wallets over the last few years has been the rise of air-gapped devices. For a lot of people, “air gap” sits at the top of their mental model of what makes a device safe. And for many threat models, isolation is genuinely valuable. It can reduce attack surface, limit remote access, and make physical tampering more obvious.
But when it comes to entropy and seed generation, air-gapping does not automatically improve your safety. In fact, it can sometimes make things harder. A design that maximizes hardware isolation may also minimize the number of potential entropy sources available at the exact moment you are creating a new wallet.
That is a dangerous tradeoff if the isolated device is the only thing generating your seed. If its internal randomness source is weak or broken, there is no second independent source to mix in. You have reduced connectivity, yes, but you have not improved the quality of the entropy that actually protects your funds.

Where things went wrong for Coldcard
The core issue came down to a major firmware change back in 2021. Coldcard had previously built on open-source Trezor code, which is one of the best-tested sets of implementations in this industry because so many people have forked it, studied it, and built on top of it. Then CoinKite moved toward its own solution, re-implemented parts of the system, and introduced a weakness in the process.
The result was that private keys generated from that point forward could have incredibly low-quality entropy. As we saw this week, those seeds could be brute-forced by anyone with suitable computing power. That is not a subtle exploit requiring insider access or physical possession of your device. It is a mathematical failure in the randomness used to create the seed.
I want to be clear: these are the kinds of problems that thorough testing should have caught in-house. Unfortunately, that did not happen. And part of the tragedy is that moving away from an open-source model can reduce the natural pressure that comes from many independent developers inspecting, forking, and stress-testing the same codebase. When fewer people build on top of a library, there are fewer eyes looking closely at it while they try to make their own product work.

This is also not a story where large language models discovered some magical, impossible backdoor. It is more mundane than that. LLMs have dramatically lowered the cost of finding and exploiting certain classes of bugs. Things that used to be too expensive or time-consuming for attackers to pursue at scale can suddenly become practical. That changes the threat model even if the underlying bug was not especially clever.
If you want the technical nitty-gritty, Block’s engineering team published an excellent write-up on what went wrong. It is a good place to go if you are technically inclined and want to understand the firmware issue in more detail.
Do I need to panic if I do not own a Coldcard?
For pretty much everyone else: no. You do not need to migrate every wallet you own this week just because of this incident. The specific bug was a very specific mistake in Coldcard firmware. It is not evidence that all hardware wallets are broken, and it is not a general reason to panic-rebuild your entire self-custody setup under stress.
That said, if you generated a seed on an affected Coldcard firmware version, updating the firmware does not fix that existing seed. The bad entropy was already baked into the seed at generation time. If you are in that position, the practical step is to generate a new seed with a trustworthy process and migrate your funds to it.
Watch out for scams after major incidents
When something like this happens, scammers go into overdrive. You should expect phishing emails that look like they come from wallet vendors. They may claim there is a vulnerability you need to check, or ask you to enter your seed phrase to verify whether it is safe. Do not do that. Ever.
You can also expect fake wallet software appearing on app stores and websites, especially if people are searching in a panic. And if someone posts a legitimate question about this incident on social media or Reddit asking whether their wallet is safe, they will probably get bombarded by private messages from scammers trying to trick them into giving away seeds or sending funds somewhere else.

The rule is simple: anything online that asks you to enter your seed words is trying to steal them. No exceptions. Do not type your seed into a website, form, chatbot, “checker,” or support tool unless it is the actual device firmware interface you trust and understand. Any legitimate low-entropy seed check would only need an address, not your full recovery phrase.
Another major scam category will be fake recovery services. You may get unsolicited emails, letters, or messages saying that funds lost in this incident have been recovered and can be returned for a fee. These are scams. They will invent excuses requiring you to pay transaction fees, taxes, “recovery fees,” or other charges before the stolen funds can be released. If someone is contacting you out of the blue about recovering your Bitcoin after an exploit like this, treat it as hostile until proven otherwise.
If you are dealing with a lost wallet, a confusing recovery situation, or want independent guidance on what options may exist, my recovery services and consultations page is available for that kind of help. But please be very careful about unsolicited “recovery” offers from strangers online.
Option 1: Use a wallet that mixes entropy well
If you are not deeply technical and all of this feels overwhelming, one reasonable answer right now is to use a wallet vendor with a long track record of doing entropy mixing properly. Trezor is an example I often point to. It has been open source for years, uses deterministic builds, and has mixed multiple sources of randomness both from the device and from your computer or phone when generating wallets.
That does not mean every user will prefer that workflow. But from a security perspective, having multiple independent entropy sources combined at seed creation time is much safer than relying on a single internal source inside an isolated device. If one source fails, you do not want that to be the only thing standing between your funds and brute force.

Option 2: Add a BIP39 passphrase, but back it up properly
Another way to add your own entropy is to use a BIP39 passphrase. This is an advanced feature, and I say that because in the recoveries I do, I see a lot of people mess this up. If you are using one, make sure you write it down correctly and test that you can restore with it.
The reason a BIP39 passphrase can help here is that your device-generated seed phrase and your passphrase are effectively mixed together to produce the actual wallet keys. In other words, even if the seed phrase generated by the device was weak or compromised in some way, adding a strong passphrase changes the resulting key material. We actually saw an illustration of this during the Coldcard incident: people who were using BIP39 passphrases, even relatively weak ones, sometimes bought themselves extra time to move funds before attackers found them.
That is not a magic shield. It is not a replacement for generating good entropy in the first place. But it can add an extra layer of protection and buy you time if something goes wrong with your device’s seed generation.
The big warning: once you use a BIP39 passphrase, it becomes part of your wallet secret. You must keep both the seed phrase and the passphrase safe. If you lose either one, you may not be able to recover the wallet. That is why I see so many recovery cases involving forgotten passphrases or faulty seed phrases. This feature can save you from some risks while creating new ones if you do not manage your backups carefully.

Option 3: Generate entropy with physical dice
The most manual option is to generate your own entropy using physical dice. This requires more work because you actually have to sit down and do it, but it has a big advantage: it is transparent. You can see what randomness went into the process. You are not trusting an opaque firmware RNG path inside a device.
In 2026, this does not always require doing everything completely by hand. Several hardware wallets now have built-in dice entropy workflows where you roll physical dice and enter the numbers into the device as you go. The device then handles the rest of the process for you. That can make the method much more practical while still giving you a visible, user-controlled source of randomness.
If you want to be especially careful, you can still do it entirely by hand if you know what you are doing. But whatever route you choose, use actual physical dice. Do not write a Python script and call it random. Do not use a phone dice app. Do not use some electronic dice gadget that came with a toy. The whole point is to introduce randomness in a way you can understand and trust.

Option 4: Build or use a DIY hardware signer
If you are really paranoid, there is also the full DIY route: assembling a hardware wallet from off-the-shelf commodity parts. This is not for everyone, but it can be attractive if you want maximum transparency over your signing device and its firmware.
One advantage in 2026 is that this has become more approachable even for people who are not deeply technical. You can point an LLM at one of these projects and ask for a quick audit or explanation. The results are generally pretty good, especially when combined with the fact that many of these open-source projects have already been audited using AI tools over the last year or so, and even more attention has gone into re-auditing them after this incident.
That does not mean you should blindly trust any DIY project. You still need to understand what you are building, where the code comes from, how it is signed, and what your threat model actually is. But for people who want to reduce reliance on a single vendor’s firmware RNG path, DIY signers can be a legitimate option.
What I would do next
If you own an affected Coldcard and generated seeds on it during the vulnerable period, do not assume updating firmware fixes old seeds. It does not. Generate a new seed using a process you trust, then migrate your funds to that new wallet. Keep your backup safe before moving anything. Do this calmly, without rushing into panic decisions or talking to random “recovery experts” who contact you first.
If you do not own a Coldcard, there is no need to tear down your entire setup right now. But use this as a reminder to think about entropy quality at seed creation time. Ask yourself: what randomness sources are actually being used? Is my device the only source? Do I have a backup that includes every part of my secret material? If I use a passphrase, do I know it works?
This incident is uncomfortable because it shows how much trust we place in devices and vendors when creating wallet secrets. A weak seed looks just like a strong one. Air-gapping can reduce some risks while reducing entropy sources. Firmware changes can introduce subtle failures that are hard to detect after the fact. And once funds are moved by an attacker, recovery is usually not something you can do yourself with a few clicks.
Stay vigilant. Be kind to people who have lost funds. Do not enter your seed phrase online. And if you want more control over your own entropy, consider mixing multiple sources, adding a properly backed-up BIP39 passphrase, using physical dice, or exploring transparent DIY signing options. The goal is not paranoia for its own sake. It is making sure the randomness that protects your Bitcoin is actually as strong as you think it is.
Stuck with a wallet recovery?
I maintain BTCRecover and have helped people recover seed phrases, passphrases and wallet passwords since 2019. Book a one-hour video session and I’ll walk you through doing the recovery yourself, without sharing your seed. Or, if you’d rather I run it, a trusted recovery is charged only if it succeeds.
Private sessions: US$100 per hour. Trusted recovery: 10% of recovered funds. I never contact people first and will never ask for your seed phrase over chat or email.
Found this useful?
These guides are free and ad-free. You can support them by sharing them, subscribing on YouTube, or sending a tip.







