, ,

Cracking Unsafe Bitcoin Dice Wallets: Coldcard Mk4 Warning

Last week, a new Coldcard Mk4 user sent 0.4 BTC to their wallet and watched it disappear within about a minute. After some back-and-forth on Reddit, they realised the problem: they had created the wallet using dice, but they had only added one roll.

That incident pulled me down a rabbit hole. I wanted to see how many other wallets might have been created with weak entropy, whether funds were still sitting in them, and whether other hardware or software platforms allowed similar low-entropy seed creation.

Watch the video for the full walkthrough, including the charts and examples I used while investigating.

What happened with the Coldcard Mk4 wallet?

The short version: the user created a new wallet using dice-only entropy, but they only entered one roll. That is far too little randomness to protect a Bitcoin wallet.

To be fair, the device did show some warning behaviour, and part of this issue was addressed in a February 2023 firmware update. But the incident still happened because the final seed had very low entropy.

Coldcard Mk4 showing dice-roll seed entry
A Coldcard Mk4 screen showing the dice-roll seed entry prompt

This is not just a theoretical problem. If an attacker can guess or brute-force the entropy used to create your seed, they can derive the same keys and sweep funds from addresses that receive Bitcoin.

Why wallets “exist” before you use them

One of the most important ideas in self-custody is that a wallet effectively exists before you ever send money to it.

When a seed is created, the possible private keys and public addresses are already determined by that entropy. If your entropy is weak, an attacker can work through the likely possibilities and find the matching address before or shortly after you use it.

This matters because attackers do not need to “hack” your hardware wallet in the usual sense. They only need to guess the seed creation process well enough to arrive at the same keys.

What I checked, and what I did not share

I used a lightly modified version of BTCRecover, an open-source recovery tool that I maintain, to check for low-entropy wallets.

BTCRecover GitHub repository
The BTCRecover GitHub repository page

Important: I am not publishing or sharing the modifications used for this kind of targeted checking. I do not want to create a how-to guide for cracking vulnerable wallets.

The goal was defensive: identify low-hanging fruit that scammers are likely exploiting, understand the scale of the problem, and see whether other platforms have similar weaknesses.

I checked several weak generation modes, including:

  1. Dice-roll based seeds up to 11 rolls.
  2. Short hexadecimal inputs up to six characters.
  3. Card-based entropy where only four cards were drawn in certain modes.

This was not meant to be comprehensive. It was a targeted check of the very low-entropy cases that can realistically be exploited quickly.

The chains checked included Bitcoin, Ethereum, Litecoin and similar Bitcoin-like chains, because many of them are easy to scan together at the address level. I did not go further into cracking weak BIP39 passphrases on top of those seeds. That was outside the scope for this investigation.

What I found: dozens of vulnerable wallets

The good news is that I found dozens of vulnerable wallets, not hundreds. The market has been quiet recently, which probably helps reduce activity in these kinds of low-entropy wallets.

Some looked like test wallets. Others were clearly not.

The bigger concern was value movement. About 22.5 BTC had passed through the vulnerable wallets I identified, and most of that value had already been taken by scammers.

Scan results showing dice-roll counts
A table of scanned wallet results with a Dice Rolls column

Most of these weak-entropy wallets were on Bitcoin. There was some value in altcoins, tokens and other assets, but the main issue was Bitcoin addresses.

The timing also stood out. Most were created fairly recently, from around mid-2021 through to the time of the video, with a noticeable spike toward late 2022 and early 2023. I suspect part of that increase may be connected to the FTX collapse and the general uncertainty around it.

Dodgy dice wallets by creation date
A chart showing insecure dice-generated wallets over time

In terms of entropy type, most appeared to have been created with dice. A smaller number used raw hexadecimal characters that are not normally available on standard dice. I also found one wallet generated using a playing-card option in a BIP39 tool, where the user had selected only a single card value.

Weak hex seed fragments highlighted
A table highlighting weak hex seed fragments

What I did with funded vulnerable wallets

When I found wallets that still contained funds, I faced an uncomfortable dilemma. I wanted to warn the owners without publishing a step-by-step cracking guide or exposing sensitive details unnecessarily.

For some funded wallets, I made a small transaction: I moved their funds out and then back in again after a couple of hours. The point was not to take anything. It was to create an obvious alert that something was wrong with the wallet, so the owner might notice and move the coins to a properly secure wallet.

Identified scammer address on a blockchain explorer
A blockchain explorer card showing an identified scammer address

This is also why I am careful about how much detail I share publicly. If you are one of the people who created a weak-entropy wallet and still have funds in it, treat this as urgent: move your coins to a wallet with strong entropy.

If you need help understanding what happened or want guidance on a recovery-related situation, my recovery services and consultations page is available. But the first step should always be to stop using a weakly generated wallet for new funds.

Why entropy matters: fewer dice rolls means less security

Entropy is just a technical way of describing randomness. In Bitcoin wallets, your seed needs enough randomness that an attacker cannot reasonably guess it.

If you create a wallet with too little entropy, the number of possible seeds becomes small enough to search. That makes the wallet vulnerable even if the hardware itself is secure.

The practical rule I want to make very clear: if you used dice to create your Bitcoin wallet and used fewer than 50 rolls, your wallet is not secure over the long term. Fifty rolls is the minimum commonly associated with a full 12-word BIP39 seed when using standard dice-based entropy.

For practical risk today:

  • Fewer than 15 dice rolls: at risk now. Move funds immediately.
  • Eight or fewer dice rolls: can be swept almost instantly by automated monitoring tools.
  • Around 12 rolls: could become much more dangerous quickly if attackers know there is value to collect.
Days to break versus number of dice rolls
A chart showing brute-force time increasing with dice rolls

The reason this changes over time is simple: computation and storage get cheaper, and attackers automate their tools. A wallet that has sat untouched for months or even a year may still be vulnerable. The fact that it has not been stolen yet does not mean it is safe.

Keyset size versus number of dice rolls
A chart showing keyset storage size increasing with dice rolls

Think of low-entropy wallets like a rising water level. As computing power improves and costs fall, more weakly protected wallets become reachable by attackers.

Coldcard Mk4 warning: dice-only seed creation

The Coldcard Mk4 is capable of several seed-generation methods, including using the onboard random number generator, using dice only, or mixing dice with the onboard entropy source.

When the Mk4 launched, one issue was that dice-only seed creation became part of the default new-seed workflow rather than being hidden in an import menu. Before a February 2023 firmware update, it could allow very low dice-roll counts with a warning that users could accept and continue.

The firmware update improved part of this by enforcing a minimum number of dice rolls during seed generation in some workflows.

Coldcard firmware changelog entry
A screenshot of the Coldcard documentation upgrade firmware page

However, as of late October 2023, I found that creating a seed through the temporary seed workflow could still allow low dice-roll counts similar to what was possible earlier in some default workflows.

Coldcard changelog highlighting minimum dice rolls
A highlighted firmware enhancement enforcing minimum dice rolls

This may be an oversight, but it matters because users can easily misunderstand the interface. The deterministic dice process includes language around mixing and shows a small hash at the bottom of the screen. Some people may have thought they were combining their dice rolls with true random entropy from the device, when in fact they were using only their dice input.

Mk3 versus Mk4 menu comparison
Two Coldcard devices comparing Mk3 and Mk4 menu layouts

This is a good example of why user interface design matters so much on advanced security devices. A feature that makes sense to an expert can become dangerous if a newer user misinterprets it.

Coldcard warning screens compared
Two Coldcard Mk4 screens comparing dice-roll and seed-vault warnings

Other hardware wallets I checked

I also went through other devices in my collection to see whether they allowed the same kind of weak-entropy wallet creation.

The devices I checked included:

  1. SeedSigner
  2. Krux
  3. Keystone
  4. BitBox02
  5. Jade

None of these devices let me create a weak, low-entropy “dodgy dice” wallet in the same way as the problematic Coldcard Mk4 workflow.

That does not mean every device is perfect for every user. It means that, for this specific failure mode, those alternatives did not expose the same dangerous path during my testing.

Software tools: Ian Coleman’s BIP39 tool

Some software tools can also create weak wallets if users enter very little entropy.

Ian Coleman’s BIP39 Mnemonic Code Converter allows you to enter a small amount of entropy and choose the seed length. It does provide warnings, but warnings are not always enough when users do not fully understand what they are doing.

Mnemonic Code Converter card-based entropy option
A BIP39 tool showing a card-based entropy input option

This is another reason why beginners should avoid manual entropy generation unless they really understand the risks.

Mnemonic Code Converter generate interface
The Mnemonic Code Converter mnemonic generation interface

User-chosen seed words increase risk

There is a growing trend of wallets that let users choose most of their recovery phrase themselves and then automatically generate the final checksum word. For example, you might pick 11 or 23 words yourself, and the software generates the last one.

That feature can be convenient, but it also increases the chance that people will create weak seeds by choosing words that “look good,” are memorable, or follow some personal pattern. If those words are not truly random, the resulting wallet may be vulnerable in much the same way as a low-dice-roll wallet.

If you choose your own words, treat them with the same seriousness as any other entropy source: they must be unpredictable.

Why use dice at all?

Many people ask why anyone would use dice instead of letting their hardware wallet generate the seed. The usual reason is distrust of the device’s random number generator. Some users worry that a bug or malicious behaviour in the onboard randomness could weaken the wallet, so they prefer to supply their own entropy physically.

That concern is understandable, but it makes seed creation an advanced task. If you do not fully understand what you are doing, you can easily create a wallet that looks secure but is actually weak.

For the vast majority of users, letting the device generate the seed is safer than manually creating entropy with dice, cards or other physical methods.

A safer alternative: BIP39 passphrase

If you do not want to trust your device’s random number generator completely, a good alternative is often to use a BIP39 passphrase.

A BIP39 passphrase adds an extra layer of security by combining the entropy from the seed with additional entropy that you choose. It can also protect your backups, because someone who finds your written seed words would still need the passphrase to access the funds.

However, passphrases add complexity. If you forget it, mistype it, or fail to back it up properly, you can lose access to everything. Every time you increase the complexity of your wallet setup, you also increase the chance of user error.

If you use a passphrase:

  1. Choose something secure but memorable enough that you can test reliably.
  2. Back it up securely and separately from the seed words.
  3. Test the full recovery process before trusting it with real funds.
  4. Do not rely on memory alone for important backups.

What you should do if you created a weak wallet

If you have ever created a Bitcoin or altcoin wallet using very little entropy, treat it as compromised until proven otherwise.

Here is what I recommend:

  1. Check how much randomness was used to create the seed.
  2. If you used fewer than 50 dice rolls for a standard BIP39 seed, assume the wallet is not safe long term.
  3. Move funds immediately to a wallet with strong entropy.
  4. Do not send large amounts to an old weak-entropy address again.
  5. Create a new wallet using your device’s onboard random number generator unless you fully understand manual entropy generation.
  6. If you use a BIP39 passphrase, back it up and test recovery carefully.

If you are unsure whether a wallet is safe, the safest action is to move the funds out and create a fresh wallet with proper entropy.

Final thoughts: simplicity is security

I want to end with a quote from Ryan Castellucchi, who did important work on brain-wallet weaknesses years ago: just because someone’s passphrase is terrible does not give you the right to steal their money.

Please do not blame the victim when people make mistakes. Pretending that users never choose weak entropy or bad passwords does not help anyone. But it also does not excuse attackers from exploiting predictable wallets.

This whole situation is a reminder not to underestimate the security value of simplicity. Trying to do things above your current technical ability can be dangerous, especially in self-custody.

Do not be fooled by marketing that says one hardware wallet or software tool is always the best option for everyone. That is never true. Different users have different needs, skill levels and threat models.

Everyone starts as a beginner. The important thing is surviving the learning curve without losing everything. If we want self-custody to be safe for more people, we need to make it easier to choose secure defaults and harder to accidentally create weak wallets.

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.

Stuck on this?

Book a one-hour session and I’ll walk you through it, or ask about a trusted recovery.

Recovery guides

More guides