Table of Contents

• You might want to know


• Main Topic


• Key Insights Table


• Afterwards...

You might want to know

Could a hardware wallet become insecure if its firmware mistakenly skips hardware randomness? How did a single implementation error lead to the rapid theft of hundreds of separate bitcoin wallets?

Main Topic

In a concentrated, rapid attack, roughly 594 bitcoin — valued at approximately $38 million at the time of the incident — were moved out of about 500 single-signature wallets in under 30 minutes. The movement occurred between 01:31 and 01:56 UTC on a Friday, concentrated within a narrow block window. The theft was traced to a flaw in key generation implemented in certain versions of Coldcard firmware introduced in March 2021.

Coldcard is a widely used, dedicated hardware wallet produced by Coinkite, a Canadian company. The device is designed to generate and protect private keys offline, keeping seed phrases and signing operations separated from internet-exposed systems. Coinkite product lines have been released in generations named Mk2, Mk3, Mk4, Q and Mk5. The risk created by this incident was not tied to the physical model number itself but to the firmware that was running when a wallet’s seed was originally created.

The core of the problem lay in randomness. A cryptographic seed phrase must be produced from a source of randomness large enough that brute-forcing or predicting the outcome is infeasible. In this case, a firmware build configuration caused Coldcard devices to bypass their onboard hardware random-number generator. A supporting library contained a check that only verified the presence of a configuration option, not whether the hardware RNG was actually enabled. When the hardware RNG was skipped, the device silently fell back to a basic software-based pseudo-random routine seeded with predictable chip data such as the device serial number and internal clock registers.

Neither the serial number nor the clock registers are secret values. A chip serial is fixed factory metadata; an attacker can obtain or guess ranges of serial values, and clock registers yield timing state an adversary can narrow down or reproduce. Because the fallback randomness derives from nonsecret values, the resulting seeds or keys became substantially more predictable than intended. Block’s Bitcoin engineering and security teams traced the problematic change to a commit dated March 1, 2021, released in Coldcard firmware 4.0.0 that same month.

Exposure extended beyond initial seed generation. The same weak generator was used to produce Coldcard paper-wallet private keys (where the generator output maps directly to the private key without further derivation), seed-splitting masks, device-cloning keys and Key Teleport transfers. In short, any feature relying on the affected random output could have produced compromised secrets.

Analysis and disclosure followed a rapid timeline. Block publicly described their findings and informed Coinkite, which acknowledged the issue. Coinkite warned users who generated seeds on Mk3 devices running firmware 4.0.1 or later to be cautious. Based on preliminary assessments, Coinkite indicated that Mk4, Q and Mk5 models do not appear affected. Block noted that it published its analysis before completing exhaustive exploitability testing because active exploitation was already underway.

The theft itself showed particular patterns: 1,324 separate bitcoin chunks were transferred across roughly 500 transactions, all within a three-block span. Of the moved funds, 562 BTC were consolidated into a single address that remained stationary at the time of reporting. Each drained wallet was single-signature and held more than 0.15 BTC; many of those wallets had been dormant for years, with balances spanning deposits from 2021 through 2026 — a distribution that matches the timeframe of the firmware mistake.

Despite the scale and visibility of the drain, the broader bitcoin market showed limited immediate reaction: prices remained around or above $64,000 during early Asian trading hours. This muted effect likely reflects a combination of market depth, the fact that the seized coins were not immediately sold on exchanges, and investors' understanding that stolen coins can remain unusable if they are tracked or otherwise difficult to monetize.

From a technical and defensive perspective, the incident reinforces several important lessons for device manufacturers, security engineers and end users. First, build-time configuration and runtime checks must be implemented with care: verifying the existence of an option is not equivalent to confirming it is active. Second, fallback mechanisms for critical operations such as cryptographic randomness should be treated with the same rigor as primary mechanisms. When a secure hardware RNG is unavailable, the fallback should either refuse to operate or use an established, auditable deterministic alternative that incorporates secure entropy externally rather than relying on predictable chip metadata.

For users, the practical takeaway is to verify the firmware used to generate any wallet seed, keep device firmware updated with vendor advisories in mind, and, if recommended by the vendor or security researchers, consider migrating funds to a new seed generated on unaffected firmware or different hardware. Because exposure depends on the firmware active when a seed was created, updating firmware alone may not remediate wallets already created with vulnerable randomness — keys already derived remain at risk and require migration.

Finally, this event underscores the broader principle that hardware devices reduce but do not eliminate the need for rigorous software engineering, secure supply-chain practices, transparent vulnerability disclosure and rapid coordinated responses when flaws are discovered. Even small teams or niche product vendors produce widely used infrastructure; mistakes at the implementation level can cascade into large-scale thefts when combined with predictable exploitation methods.

Key Insights Table

Afterwards...

Looking forward, the incident should prompt vendors to strengthen build and runtime validation, adopt safer fallback policies for critical cryptographic operations, and improve vulnerability disclosure procedures. Users should audit firmware provenance for devices used to create seeds, follow vendor advisories closely, and assume that any seed created on a device with the affected firmware may need migration. The event highlights that cryptographic security depends not only on hardware capabilities but also on careful software engineering, testing and responsible coordination between researchers and manufacturers to limit harm.