Coldcard Seed Vulnerability: What Happened, Who's Affected, How to Fix It

Last updated: 2026-08-11. This page will be revised as the story develops — see "How this page stays current" below.

What happened

Starting the night of July 30, 2026, attackers began systematically draining Bitcoin from Coldcard wallets whose seeds were generated with weak randomness — not through phishing, not through malware, and not by physically touching the device. By early August, independent blockchain-analytics firms including TRM Labs and Galaxy Research had confirmed thousands of Bitcoin drained from thousands of addresses across multiple sweeps, some emptied within minutes of exposure. Reported totals vary by source and by the moment each was published — TRM Labs' original figure was roughly 1,800 BTC (~$116 million) from over 5,200 addresses; later reporting from outlets including TechCrunch put the figure above $130 million as more affected wallets were identified. This is one of the largest hardware-wallet failures in Bitcoin's history, and coverage broadly describes it as one of the largest single crypto thefts of 2026.

What actually went wrong, mechanically

A March 2021 firmware update (starting at version 4.0.0/4.0.1) introduced a build error in a cryptography library Coldcard relies on. The code checked whether a hardware-randomness setting existed rather than whether it was switched on — and since Coinkite had it deliberately set to off (they use their own separate hardware-randomness wrapper), the firmware silently fell back to a much weaker software substitute, seeded mainly from the device's serial number and boot timing rather than genuine randomness. The result: seeds that displayed completely normally on-screen but were dramatically easier to guess than intended — down to roughly 40 bits of effective security on Mk2/Mk3 (a range a laptop can brute-force in hours) and roughly 72 bits on Mk4/Mk5/Q (still far below the 128 bits the design calls for). The flaw sat undetected in shipping firmware for over five years because the device's real hardware random-number generator was present and working — this one code path just never called it.

Affected firmware versions

This table is sourced from Coinkite's own advisory, the authoritative source — always check it directly if you're making a decision based on firmware version, since it may be updated after this page was last revised.

What actually stayed safe

As of the most recent tally, no genuine multisig wallet or Taproot address had been drained through this bug alone — but that protection is conditional, not automatic: a multisig setup is only safe if not every key in the quorum came from a vulnerable Coldcard generating its own entropy. Two other protections fully neutralize the bug on their own: a seed created with at least 50 independent, private dice rolls, or a wallet protected by a strong, unique BIP-39 passphrase (a second secret layered on top of the 24 words, producing an entirely separate wallet an attacker can't reach just by guessing the weak seed). All three were optional steps the standard setup flow allowed skipping — which is exactly why this mattered at scale.

What to do if you're at risk

Updating firmware is necessary, but it does not fix an already-generated seed. The patch stops the device from creating new weak keys — it cannot repair a key that already exists, and moving that same seed to a different wallet or device doesn't help either, since the weakness lives in the seed itself, not the hardware. If you generated a Coldcard seed before installing patched firmware, and didn't use 50+ dice rolls, a strong passphrase, or genuine independent-entropy multisig, treat that seed as compromised:

1. Install the current patched firmware for your device (see the table above, and verify against Coinkite's own advisory).
2. Generate a completely new seed on the patched firmware — do not reuse or "refresh" the old one.
3. Move all funds from the old seed to the new one as soon as practical.
4. Never fund the old seed again, even after moving funds out of it.

Coinkite's CEO, Rodolfo Novak, confirmed this guidance publicly and took responsibility for the bug. One more warning, because scammers move fast after news like this: never type your seed phrase into any website claiming to "check" whether you're affected. There is no legitimate online checker that needs your seed phrase — the only way to know is your device's own firmware version and how the seed was originally created.

Not the first of its kind

Weak randomness has quietly undermined Bitcoin wallets before — the 2023 "Milk Sad" bug and the tool behind the 2022 Wintermute hack were both, underneath, the same category of failure: predictable randomness that looked identical to real randomness until someone checked. That's a real point in favor of open-source, auditable hardware — bugs like this are eventually findable in public code — but it's also a reminder that "open source" catches problems eventually, not instantly, and doesn't substitute for the extra protections (dice rolls, a passphrase, real multisig) described above.

How this page stays current

This is a standalone status page, not a news article with a fixed publish date — it gets revised as new, sourced information becomes available (a firmware update, a revised loss total, a clearer remediation step). It summarizes and links to primary sources rather than replacing them; if anything here conflicts with Coinkite's own advisory or a current tracker like TRM Labs', defer to those. For a factual, non-hype comparison of Coldcard against other hardware wallets (including ones without this issue), see this site's buyer's guide.