Hardware wallets have long been considered the most secure way to store digital currency, but a recent incident involving devices from the company Coinkite will make many reconsider this opinion. On July 30, 2026, malicious actors withdrew 594.5 bitcoins (approximately $40 million) from five hundred addresses in just twenty-five minutes.
The Nature of the Software Bug in Simple Terms
The root of the problem lies in a bug in the software code that went unnoticed for a full five years. The Coldcard hardware wallet was supposed to create a seed phrase using a specialized secure chip that generates truly random numbers. However, due to a typo in one macro, Coinkite developers accidentally disabled this function back in March 2021.
This gross violation of industry standards led to the device creating keys based on predictable data instead of relying on secure protection: the CPU's serial number and readings from the internal system clock. The generated words looked like a standard secure set, but due to the predictability of their creation process, hackers could easily brute-force them. The level of cryptographic strength turned out to be catastrophically low: for the Mk2 and Mk3 models, the search space was reduced to approximately 40 bits, while in the newer Mk4, Mk5, and Q versions, the final entropy amounted to about 72 bits instead of the required 128 bits.
Attack Mechanics and Company Response
The attackers did not need physical access to the devices or phishing links. They simply took the parameters of this "leaky" generator and organized an offline brute-force attack. By generating millions of possible seed phrases, the hacker checked their presence in the public ledger, found addresses with a balance, and independently signed transactions to transfer the funds.
Initially, Coinkite stated that new versions of the hardware were not at risk. However, subsequent analysis forced management to admit that all devices running compromised firmware were vulnerable. Coinkite CEO Rodolphe Novak, known by the pseudonym NVK, apologized on social media X for the mistake, confirming the developers' full responsibility but ruling out any possibility of financial compensation for affected users.
Required Actions for Owners
It is important to understand: simply installing a fresh firmware update does not eliminate the threat for an already created seed phrase. The cryptographic key has been compromised since the moment of its generation. To protect assets, it is necessary to follow a strict action algorithm:
-
Update the device to secure software versions: 4.2.0 for Mk3, 5.6.0 for Mk4 and Mk5, or 1.5.0Q for the Q model.
-
Generate a new seed phrase directly on the updated device.
-
Completely transfer all funds to new addresses created using this fresh seed phrase.
Using a secure passphrase according to the BIP-39 standard can also reduce risks, but this does not negate the need to migrate to a new key. The TAPSIGNER, OPENDIME, and SATSCARD products are not affected by this incident as they operate on different software bases.
Coinkite representatives noted that the open-source nature of the code theoretically allowed for the detection of this vulnerability using artificial intelligence tools, although the company's internal checks shortly before the incident did not reveal the problem. Some analytical reports, including data from Galaxy Research, suggest that the total volume of withdrawn funds could reach 1,000 bitcoins.
The situation demonstrates that even specialized hardware requires constant independent code auditing, especially regarding cryptographic functions.
AI Opinion
From the perspective of machine data analysis, the Coldcard incident is not the first case where a random number generator defect remains unnoticed for years. A similar scenario unfolded with the OpenSSL package in Debian: a bug in a 2006 patch narrowed the key space to 32,767 options, and the vulnerability was only discovered two years later, jeopardizing SSH and SSL keys worldwide.
The macroeconomic aspect of the story also deserves attention: trust in hardware wallets was built precisely on the fact that a closed physical medium is more reliable than software solutions, and this incident blurs that boundary. Open-source code theoretically speeds up the detection of such errors through independent audits and automated tools, but five years of silence show that theory and practice diverge. The question remains: will algorithmic code checks ever be able to fully replace human audits of critical cryptography?
