Fault Injection Attack Recovers $2M From Trezor One After Total Credential Loss
SurvivedWallet passphrase was unavailable — a recovery path existed and access was restored.
A Bitcoin holder with over $2 million stored on a Trezor One hardware wallet lost access to the device after forgetting both the PIN and seed phrase. Without these credentials, the funds appeared permanently inaccessible through normal recovery procedures.
Security researcher Joe Grand was engaged to investigate potential recovery options. Grand identified a critical vulnerability in the wallet's firmware version: during firmware update operations, the device copied sensitive credentials to RAM rather than securely moving them. Crucially, these credentials persisted in the device's flash storage after the update completed, creating a window of exposure.
Grand employed a fault injection attack—a technique that deliberately induces computational errors in the device's processor to bypass security controls. By carefully triggering faults while the wallet was powered and operational, he was able to read the wallet's RAM and extract both the PIN and seed phrase in plaintext.
With credentials recovered, Grand unlocked the wallet and restored the owner's access to the $2 million in Bitcoin. The recovery succeeded due to a convergence of factors: the device's age and specific firmware version, the particular credential-handling flaw during updates, and Grand's deep expertise in hardware security exploitation and chip-level attacks.
This case illustrates a fundamental vulnerability in early hardware wallet firmware design—the failure to properly segregate volatile and persistent memory during firmware updates. It also demonstrates that physical access combined with advanced hardware hacking techniques can sometimes circumvent even lost-credential scenarios, though such recovery remains expensive, time-consuming, and accessible only to specialists.
| Stress condition | Passphrase unavailable |
| Custody system | Hardware wallet (single key) |
| Outcome | Survived |
| Documentation | Present and interpretable |
Why passphrases fail years after they are set
The failure mode documented consistently across observed cases is temporal: the passphrase is set with confidence, not used for an extended period, and then cannot be reproduced exactly when needed. A single character difference — different capitalization, an added space, a slightly different special character — produces a different wallet with a zero balance. The holder may be certain they remember the passphrase while being unable to produce the exact string that was originally set.
What makes this particularly difficult is that there is no signal at the moment of failure. A wrong passphrase does not produce an error message. It opens an empty wallet. The holder sees a zero balance and typically concludes the passphrase was wrong — but without knowing which part was wrong, or by how much.
Professional passphrase recovery services can attempt permutations when the holder has partial information: they remember the general structure, typical patterns they use for passwords, the approximate length, or that it included a specific word. Recovery from total non-recollection is not feasible.
The preventive action is to store a passphrase record — not with the seed phrase, which would defeat its security purpose, but in a separate secure location accessible to the holder and potentially a designated recovery person. A passphrase that exists only in memory has a time horizon: it will eventually be forgotten, and the timing is unpredictable.
Translate