Coldcard Hacked for $70M: How Do You Keep Bitcoin Safe if Cold Wallets Can Be Hacked?

Bitcoin held steady while Coldcard users lost roughly $70M

Galaxy says 1,196 addresses were drained and roughly 1,082 BTC disappeared, yet Bitcoin stayed rangebound anyway. That suggests the incident was not a break in Bitcoin itself and not proof that cold wallets are useless. The failure happened at key generation, when some seeds were created with weak entropy.

If analysts are right, those seeds were never truly random. That meant the private keys could, in some cases, be recomputed by an attacker. This was not a network exploit or a broken blockchain; it was a flawed origin for certain keys.

The other important detail is that only single-signature wallets were emptied. That does not make single-signature self-custody unsafe by default, but it does show why key generation, passphrase use, and multisig design matter so much.

What actually broke in Coldcard firmware

The hardware RNG was bypassed during seed generation

A hardware wallet marketed on hardware-isolated key generation ended up, on certain firmware versions, with that path wasn't actually being used for seed randomness. Coinkite said a firmware bug caused seed generation to fall back to a weak software PRNG. In practical terms, seeds that should have been extremely hard to guess became guessable ones.

That makes this less of a dramatic "device was hacked" story and more of an entropy problem at the moment the wallet was created.

The 25-minute sweep shows why weak entropy is so dangerous

The speed of the theft does most of the explaining. Attackers moved a large share of the stolen coins in a under 25 minutes sweep across hundreds of wallets. That is consistent with bulk key recomputation, not with someone trying seeds one by one.

The issue spread beyond the original Mk3 warning

Coldcard first flagged Mk3, then expanded precautions to Mk4, Mk5, and Q. Fixed firmware is already available, including 5.6.0 or later for Mk4/Mk5 and 1.5.0Q or later for Q. But updating firmware does not fix seeds that were already created under the faulty process.

Coinkite's warning is explicit: funds may still be at risk if the seed was not created with at least 50 independent, private dice rolls and the wallet was not protected by a strong, unique BIP-39 passphrase. Even then, a strong passphrase does not repair an affected seed. The practical takeaway is to treat this as a seed-migration problem, not just a firmware-update problem.

What actually protects Bitcoin in self-custody

Entropy comes before isolation

Air-gapped designs help, but they do not compensate for bad seed generation. When randomness fails at creation, the weak point is still where the seed was born. The Coldcard case shows why the order of defense matters:

1. Entropy at creation is the foundation.
2. A strong, unique passphrase can reduce exposure, but it is not a cure-all.
3. Multisig limits the blast radius when one key or one backup is compromised.
4. Firmware hygiene matters because software mistakes can still undermine a hardware wallet.

What users should do next

Coinkite's guidance is straightforward: if your seed was not created with at least 50 independent, private dice rolls, you should treat it as potentially vulnerable. A passphrase helps only if it is strong, unique, secret, and separate from the seed backup. And even then, a strong passphrase does not repair an affected seed.

If you hold meaningful Bitcoin, the key question is simple: are you treating this as a software patch, or as a reason to migrate weak seeds to a fresh, carefully generated backup? In cases like this, user behavior usually matters as much as the device itself.