The Exploit: Root Cause and Impact
Published 8/2/2026, 10:22:09 AM
The Coldcard exploit, which culminated in a $70.2 million (1,082.65 BTC) theft on July 30, 2026, was a firmware-specific implementation failure rather than a fundamental flaw in hardware wallet architecture. While the bug itself was unique to Coldcard's codebase, it has exposed three systemic industry risks regarding how hardware wallets generate and verify entropy.
The Exploit: Root Cause and Impact
The vulnerability originated from a firmware configuration error introduced in March 2021 (v4.0.0). A build-time flag (MICROPY_HW_ENABLE_RNG) was incorrectly set to zero, causing the device to bypass its hardware random number generator (RNG) and instead use a predictable software fallback (MicroPython's Yasmarang PRNG) [Source: https://engineering.block.xyz/blog/predictable-rng-fallback-and-32-bit-reseed-in-coldcard-firmware].
| Metric | Details |
|---|---|
| Total Stolen | ~1,082.65 BTC from 1,196 addresses [Source: https://blog.coinkite.com/coldcard-mk3-seed-generation-warning/] |
| Vulnerability Window | March 2021 – July 2026 |
| Effective Entropy | Mk3: ~40 bits; Mk4/Mk5/Q: ~72 bits (Standard is 128 bits) [Source: https://blog.coinkite.com/entropy-technical-backgrounder/] |
| Attack Duration | 41 minutes (automated sweep) |
| Root Cause | Library checked if a macro existed rather than if it was enabled [Source: https://engineering.block.xyz/blog/predictable-rng-fallback-and-32-bit-reseed-in-coldcard-firmware] |
Does this expose a broader hardware wallet flaw?
Security researchers from Block and Galaxy Research characterize this as an isolated implementation error rather than a flaw in the BIP39 standard or secure element hardware. However, it highlights three broader vulnerabilities inherent to the hardware wallet ecosystem:
- The "Entropy Blind Spot": Users currently have no mechanism to verify if the 12 or 24 words generated by any hardware wallet are truly random. This "black box" generation is a systemic risk across all manufacturers.
- Firmware Complexity Risk: As wallets add features like NFC, Bluetooth, and Python-based environments, the surface area for "silent" bugs increases. In this case, a simple build flag error remained undetected for over five years.
- The Update Fallacy: A critical realization for the ecosystem is that firmware updates do not secure existing seeds. Because the flaw existed at the moment of seed generation, any wallet created between 2021 and 2026 remains compromised even after the device is patched [Source: https://blog.coinkite.com/coldcard-mk3-seed-generation-warning/].
Affected vs. Unaffected Devices
The vulnerability was strictly limited to the Coldcard firmware codebase. Other manufacturers and even other Coinkite products were not impacted.
- Affected: Coldcard Mk2, Mk3, Mk4, Mk5, and Coldcard Q running firmware released between March 2021 and July 2026.
- Unaffected: TAPSIGNER, OPENDIME, and SATSCARD (which use different codebases) [Source: https://blog.coinkite.com/coldcard-mk3-seed-generation-warning/].
- Safe Exceptions: Users who generated their seeds using 50+ physical dice rolls were unaffected, as this process bypasses the internal RNG entirely [Source: https://blog.coinkite.com/entropy-technical-backgrounder/].
Security Implications and Mitigations
The exploit underscores the importance of external entropy and defense-in-depth. Users who utilized a strong BIP-39 passphrase were significantly better protected, as the passphrase adds a layer of entropy that the attacker could not predict via the RNG flaw. Similarly, multisig setups were only compromised if all participating keys were generated on affected Coldcard devices during the vulnerable period.
Conclusion: The exploit was a specific firmware bug, but it serves as a warning that the industry lacks a standardized way for users to audit the randomness of their keys at the time of creation. For affected users, the only resolution is to generate a new seed on patched firmware and migrate funds immediately.