Suspected implant in a Trezor Safe 5 sold by CryptoBilis
One reported unit, not yet examined by Trezor at the time of writing. The report concerns how a device reached a buyer, not the device design.
Every recorded Trezor security incident, newest first, with severity, fix status and sources.
One reported unit, not yet examined by Trezor at the time of writing. The report concerns how a device reached a buyer, not the device design.
Names, contact details and shipping addresses of up to 80,689 Trezor customers were exposed through ShipMonk, raising the risk of targeted phishing. Trezor states its systems and devices were not affected.
A lab attacker holding a Trezor Safe 7 can run arbitrary firmware on its TROPIC01 secure element and, per Tropic Square and Ledger, compromise its MAC-and-Destroy protection. Trezor states this alone does not expose the PIN, funds or wallet backup and that no user action is needed.
An attacker with lab equipment could turn a Safe 3 into a device running malicious firmware that still passes authenticity checks, then sell it to a victim. Devices bought from official sources and the Safe 5 are not affected, according to Trezor.
An attacker with physical access and laboratory equipment could, per NinjaLab, extract ECDSA keys from the OPTIGA Trust M chip used in the Trezor Safe 3 and Safe 5. Trezor states the recovery seed and funds are not at risk because that cryptography does not protect the wallet backup.
Names and email addresses of up to 66,000 support contacts were exposed, and some then received emails asking for their recovery seed. Trezor states no funds were compromised and the devices were unaffected.
Anyone who set up a wallet on such a counterfeit used a seed the attacker already knew. Owners of genuine devices bought through official channels are not affected.
Owners who installed the fake Trezor Suite and entered their recovery phrase gave attackers control of their funds. Subscribers' email addresses and, in some cases, names and IP addresses were exposed, according to Trezor, which invites further targeted phishing.
A Trezor user signing on a compromised computer could be led to sign twice and lose most of the transaction's inputs to fees. Trezor lists the issue as fixed in Trezor One 1.9.1 and Model T 2.3.1.
Someone holding a Trezor One or Model T without a passphrase could recover its seed with specialised equipment in about 15 minutes. A device protected by a BIP39 passphrase is not exposed, according to both Kraken and Trezor.
Anyone holding a Trezor One or Model T without a strong passphrase could, according to Ledger, extract its seed in minutes with inexpensive equipment. Trezor states a passphrase fully mitigates the attack and disputes that it was disclosed to the company.
An attacker with physical access to a Trezor One on firmware older than 1.8.0 could read the seed from memory unless a passphrase was set. Trezor fixed them in Trezor One firmware 1.8.0; the Model T was not directly affected and received the same protections in 2.1.0.
Someone with physical access to an opened Trezor One running firmware older than 1.5.2 could read its seed and PIN from RAM; passphrase-protected wallets were not exposed, according to the researcher's write-up. Trezor states the issue was fixed in 2017.
More entries do not mean a less secure brand: widely sold devices draw more published research and more targeted attacks, and some vendors disclose more than others.
WalletInsights. Hardware wallet security incident tracker. Last checked November 3, 2026. https://walletinsights.io/en/security-incidents/trezor/
Vendors, researchers and readers can send corrections or statements to security@walletinsights.io. We check them against the sources, update the entry and record the change in its log.