Kraken extracts Trezor One and Model T seeds by voltage glitching

On 31 January 2020 Kraken Security Labs published research showing it could extract the encrypted seed from Trezor One and Trezor Model T devices with about 15 minutes of physical access. Kraken says it disclosed the full details to Trezor on 30 October 2019.

According to Kraken, the attack uses voltage glitching to bypass the microcontroller’s read protection and dump the flash memory; the PIN protecting the encrypted seed is then brute-forced, which Kraken reports took under 2 minutes for a 4-digit PIN on a single CPU core. Its setup cost several hundred dollars, and it estimated a consumer-friendly glitching device could be mass-produced for about $75. Decrypt notes that the attack includes opening the device’s case.

Kraken attributes the flaw to the microcontroller and says fixing it would require new hardware; its CSO told The Block that Trezor “would essentially need to put out a new device”. In its response of 31 January 2020, Trezor states that the attack requires opening the case, removing the chip from the board and connecting it to specialised equipment, so an attacked device shows visible signs of tampering, and that a strong passphrase fully mitigates it.

What to do:

• Enable a BIP39 passphrase. It is not stored on the device, so it blocks this attack (Kraken, Trezor).
• Do not let anyone gain physical access to your device.