Coldcard Wallet Vulnerability: Who Is Affected and Who Should Not Panic?

A recent COLDCARD vulnerability appears to have weakened the randomness used to create some Bitcoin wallet seeds, and affected users may need to move funds to a fresh seed. This does not mean all cold wallets are compromised, and it does not mean every COLDCARD user has the same level of risk.

If you use a hardware wallet, the first thing to understand is this: the risk is tied to how and where the seed phrase was originally created. A firmware update can fix future seed generation, but it cannot go back in time and make an old weak seed stronger.

What happened with COLDCARD?

Coinkite, the company behind COLDCARD, published a security advisory warning that some seeds generated on COLDCARD devices may be at risk because of a seed-generation entropy issue. In plain English, the device may not have used as much true randomness as it was supposed to when creating some wallet seeds.

A Bitcoin wallet seed phrase is supposed to be so random that guessing it is practically impossible. If the randomness is weak, the number of possible seeds can shrink from an impossible search to a much smaller search. That is a serious problem because the seed phrase is the root of the wallet. If someone can recreate or guess the seed, they can derive the private keys and move the funds.

Around the same time, researchers and on-chain analysts reported a coordinated sweep of roughly 594 BTC, worth about $38 million at the time, from around 500 single-signature wallets. Public reporting has linked the timing and wallet patterns to the COLDCARD issue, but there is an important caveat: Coinkite has said its investigation is continuing, and not every public claim about the theft should be treated as settled fact.

Still, the risk is serious enough that both Coinkite and outside researchers have urged affected users to migrate carefully.

Which COLDCARD users are affected?

The main question is not simply, “Do you own a COLDCARD?” The better question is, “Was your seed generated on affected COLDCARD firmware?”

According to Coinkite’s July 31 update:

• COLDCARD Mk3 seeds generated on firmware version 4.0.1 or later may be at risk.
• COLDCARD Mk4 and Mk5 seeds generated before firmware version 5.6.0 are also affected, though Coinkite says the impact is less severe than Mk3.
• COLDCARD Q seeds generated before firmware version 1.5.0Q are also affected.
• Updating firmware helps future seed generation, but it does not repair a seed that already exists.

Block’s Bitcoin Engineering and Security team published a separate technical analysis. Block traced the issue to an RNG integration problem involving ngu.random, MicroPython’s fallback random generator, and a hardware RNG configuration check that did not behave as intended. Block’s affected-device table is slightly different in wording and version detail from Coinkite’s advisory, but the practical guidance is similar: exposure depends on the firmware used when the secret was generated.

That distinction matters. If you generated a seed years ago on affected firmware and later restored that same seed into a different wallet, the seed is still affected. Moving a weak seed into a different device does not make the seed stronger.

Who is at highest risk?

The highest-risk profile appears to be a single-signature wallet where the seed was generated by affected COLDCARD firmware, with no strong BIP39 passphrase, no private dice entropy, and no multisig protection.

That means a user who did something like this may be in the danger zone:

1. Bought or used a COLDCARD Mk3 after the affected firmware period began.
2. Created a normal new wallet seed on the device.
3. Did not add independent dice rolls during seed creation.
4. Did not use a strong, unique BIP39 passphrase.
5. Used the seed for a single-signature wallet.
6. Still has funds controlled by that seed.

This does not mean every person in that group has already lost funds. It means the setup has the kind of exposure that the advisory is warning about.

If that describes you, this is not the time to get creative. Read the official Coinkite advisory, confirm your device and firmware situation, update if applicable, generate a new seed only on fixed firmware or unaffected hardware, test the new wallet with a small transaction, and then migrate carefully.

Who should be less worried?

Some users should still pay attention, but they have less reason to panic.

You are likely in a better position if:

• Your seed was not generated on a COLDCARD.
• You use a different hardware wallet and never imported a COLDCARD-generated seed.
• You use TAPSIGNER, OPENDIME, or SATSCARD. Coinkite says those products use different codebases and are not affected by this bug.
• You generated the seed with enough private dice entropy and used the final dice-added seed words.
• You used a strong, unique BIP39 passphrase that was never typed into a website, phone, or untrusted computer.
• You use multisig with independent keys from different devices or vendors, and the affected COLDCARD was only one signer.
• The affected wallet is empty and you have already moved funds to a fresh seed generated safely elsewhere.

There are caveats here. A weak passphrase may not help much. A passphrase is not the same thing as a COLDCARD PIN. A multisig setup is only as good as its actual threshold and backup process. Dice rolls only help if they were fair, private, entered correctly, and used to produce the final wallet seed.

So, “less worried” does not mean “ignore the advisory.” It means your risk may be much lower than someone who relied only on the affected device’s normal seed-generation path.

Why not all cold wallets are affected

This incident is about a COLDCARD firmware and seed-generation issue. It is not a universal cold wallet failure.

A cold wallet is a broad category. It can mean a hardware wallet, an air-gapped signing device, a paper backup, a multisig setup, or another offline custody method. Different products use different firmware, chips, libraries, build systems, entropy designs, and backup flows.

So if someone says, “Cold wallets got hacked,” that is too broad. The more accurate statement is that certain COLDCARD-generated seeds may have been created with weaker randomness than intended.

That difference matters because people may overreact in the wrong direction. The lesson is not to abandon cold storage. The lesson is to understand that cold storage still depends on correct seed generation, verified firmware, careful backups, and sane migration procedures.

Air-gapping helps reduce some attack paths. It does not magically fix a weak seed that was created at the beginning.

What was the technical vulnerability?

The technical issue centers on randomness.

When a wallet creates a seed phrase, it needs high-quality entropy. Entropy is the unpredictable input used to create the seed. For a Bitcoin wallet, weak entropy is dangerous because the seed phrase controls the private keys.

Block’s analysis says COLDCARD firmware contained an RNG integration error. The firmware expected to use secure hardware randomness, but a library path caused ngu.random to bind to MicroPython’s deterministic fallback generator instead of the intended hardware RNG.

The short version is this:

• COLDCARD had its own hardware RNG wrapper.
• A configuration macro related to MicroPython’s hardware RNG was defined as zero.
• The library checked whether the macro existed, rather than whether it was enabled.
• Because the macro existed, the build passed.
• But the random function could fall back to a deterministic software generator.

For older affected Mk2/Mk3 firmware paths, Block says no cryptographic entropy was added to that ngu.random path. For Mk4, Q, and Mk5, Block says secure-element entropy was added, but only four bytes were retained for reseeding, leaving at most a 32-bit securely distinguished contribution in that part of the design. Coinkite’s advisory describes the newer-device impact as about 72 bits of entropy rather than the expected 128 bits.

Do not get lost in the numbers if you are a normal user. The practical point is simple: a seed that should have been effectively impossible to search may have been far easier to search for affected setups.

How was the vulnerability discovered?

Public information suggests the vulnerability came to light after users and researchers noticed suspicious wallet drains, then investigators traced the problem back through the firmware.

According to Block, on July 30, 2026, Block and other security researchers noticed reports of COLDCARD users losing funds. Block says it began investigating, independently found the root-cause vulnerability, shared findings with Coinkite, and published early because it believed active exploitation was underway.

On-chain analysts also reported a coordinated sweep pattern: hundreds of single-signature addresses, many apparently dormant for years, moved in a tight window. That kind of pattern is different from a normal phishing incident where one person types a seed into a fake website. It suggested to several researchers that the private keys or seeds may have been weak from creation.

Coinkite has also reportedly said it has to assume someone used AI to review older firmware and find the flaw. That claim is difficult to prove from the outside. It is plausible in the sense that open-source firmware can be reviewed by anyone, and modern AI tools can help search code for suspicious patterns. But until Coinkite publishes a full technical post-mortem with evidence, the AI-discovery angle should be treated as an informed suspicion, not a proven fact.

What is clearer is the investigation path: reports of theft, on-chain pattern analysis, firmware review, root-cause analysis, emergency advisory, and firmware hotfixes.

How could the vulnerability be exploited?

The likely exploitation path does not require the attacker to physically steal your COLDCARD.

If the seed-generation process was predictable enough, an attacker could work offline. They could generate candidate seeds, derive wallet addresses from those candidates, and compare them against public Bitcoin addresses or xpub-derived addresses. Once a candidate seed produced an address with real funds, the attacker could derive the private key and spend the Bitcoin.

This is why public addresses matter in this kind of attack. A Bitcoin address is not supposed to reveal your private key. But if the seed space is much smaller than expected, the address becomes a way to check guesses. It is like trying keys in a lock, except the attacker does the trial work with computers and public blockchain data.

That also explains why updating firmware is not enough. The weak seed is already the root of the wallet. A firmware update can stop the device from making weak seeds in the future. It cannot rewrite the old seed controlling old addresses.

What should affected users do?

Follow the official advisory first. Do not rely on random social media threads for the final migration process.

The general safe path looks like this:

1. Confirm your COLDCARD model and firmware.
2. Read Coinkite’s current advisory and firmware upgrade instructions.
3. If using Mk4, Mk5, or Q, update to the fixed firmware before generating a new seed.
4. Generate a completely new seed. Do not reuse or restore the old seed as the replacement.
5. Record and verify the new backup.
6. Verify a receive address on the device screen.
7. Send a small test transaction first.
8. Confirm the test arrived and that you can still recover/sign properly.
9. Move the remaining funds.
10. Keep the old backup until the migration is fully confirmed.

If the affected Mk3 is your only available device, Coinkite describes interim options involving a strong BIP39 passphrase or an advanced dice-only seed path. Those steps can help, but they require care. A typo in a passphrase creates a different wallet. A lost passphrase can permanently lock you out. A badly handled migration can create a new loss while trying to avoid the original risk.

Slow is smooth. Smooth is safer.

What should users avoid doing?

Do not type your seed phrase into any website claiming to check whether you are affected.

Do not enter your BIP39 passphrase into a random phone app, web tool, or desktop program unless you already fully trust that environment and understand the consequences.

Do not assume that moving the same old seed into another hardware wallet fixes the problem.

Do not rush a full-balance transfer before verifying the new wallet with a small test.

Do not throw away the old backup until you are completely sure the new wallet is working and the funds have moved.

And do not assume this is a reason to abandon self-custody altogether. Self-custody has real responsibilities. One of those responsibilities is responding calmly when a tool you used has a serious advisory.

What does this mean for hardware wallet trust?

The hard lesson is that hardware wallets are not magic boxes. They are small computers running firmware. Good hardware wallet design matters, but so do code review, entropy design, reproducible builds, independent audits, update channels, user education, and emergency migration instructions.

Open-source firmware helps because researchers can inspect it. Signed firmware helps because users can verify that they are installing an approved release. Air-gapping helps because it reduces live connection risk.

Each of those controls still has limits.

A signed firmware release can contain a bug. An air-gapped wallet can create a weak seed. Open source can be reviewed by good researchers and bad attackers. None of that makes hardware wallets useless. It means serious custody requires layered controls rather than blind trust in a brand name.

For larger balances, this incident strengthens the case for:

• strong BIP39 passphrases,
• private dice entropy during seed creation,
• multisig with independent signers,
• multi-vendor setups,
• tested recovery procedures,
• and staying subscribed to manufacturer security advisories.

You do not need every advanced setup for every small balance. But the bigger the balance, the less comfortable you should be with one device, one seed, one signer, and no extra protection.

FAQ

Are all cold wallets affected by the COLDCARD vulnerability?

No. This advisory concerns specific COLDCARD seed-generation paths and firmware versions. Other cold wallets are not automatically affected unless they use a seed that was originally generated on affected COLDCARD firmware.

Are Trezor, Ledger, Bitkey, and other hardware wallets affected?

Based on the sources reviewed for this draft, the issue is specific to COLDCARD firmware and seed generation. Other wallet brands are not affected by this COLDCARD bug unless a user imported a weak COLDCARD-generated seed into them.

Does updating COLDCARD firmware fix my old seed?

No. Firmware updates can fix future seed generation, but they do not change an existing seed. If your current seed was created with weak entropy, the safer long-term move is to generate a fresh seed on fixed firmware or unaffected hardware and migrate funds.

Is a BIP39 passphrase the same as my COLDCARD PIN?

No. The PIN unlocks the device. A BIP39 passphrase is an additional secret that changes the wallet derived from the seed phrase. If you lose the passphrase, you may lose access to that passphrase-protected wallet.

If I used dice rolls, am I safe?

Coinkite says at least 50 fair, independent, private dice rolls can provide at least 128 bits of entropy, and 99 or more can provide about 256 bits. If you are unsure how many rolls you used, whether they were private, or whether the final seed was the dice-added seed, follow migration guidance.

Should I panic-transfer everything right now?

A careless migration can cause its own loss. Affected users should act with urgency, but the process should be deliberate: verify firmware, generate a fresh seed, verify backups, test with a small transaction, and then move the larger balance.

Final thoughts

The COLDCARD incident is serious because it touches the root of self-custody: seed generation. If the seed is weak, the wallet can be weak no matter how careful the user was afterward.

But this should not be flattened into “cold wallets are unsafe.” That would be the wrong lesson. The better lesson is that cold storage is a system. The device matters. The firmware matters. The seed-generation method matters. The passphrase matters. The backup matters. The recovery plan matters.

If you are affected, move carefully. If you are not affected, use this as a reminder to review your own setup before a crisis forces you to do it under pressure.

Sources reviewed:

• Coinkite, “Mk3 Security Advisory,” updated July 31, 2026: https://blog.coinkite.com/coldcard-mk3-seed-generation-warning/
• COLDCARD firmware upgrade documentation, updated July 31, 2026: https://coldcard.com/docs/upgrade/
• Block Engineering, “Predictable RNG Fallback and 32-Bit Reseed in COLDCARD Firmware”: https://engineering.block.xyz/blog/predictable-rng-fallback-and-32-bit-reseed-in-coldcard-firmware
• COLDCARD security and verification documentation: https://coldcard.com/resources/security/coldcard-security-and-verification
• Public reports from Cointelegraph, Decrypt, CryptoSlate, and other outlets were used only as supporting context for the reported 594 BTC sweep and public timeline.

Disclaimer: The content on this blog is provided for general informational and educational purposes only and does not constitute financial, investment, legal, or tax advice. Cryptocurrency and Bitcoin investments are highly volatile and involve substantial risk. Mitigate your own risk with your own management practices. Always conduct your own research and consult a qualified professional before making financial decisions. Past performance does not guarantee future results. The author and blog assume no responsibility for losses arising from reliance on this content. Any opinions expressed are the author’s own, and any holdings, sponsorships, or affiliate relationships are disclosed where applicable.