Cifrado de hardware en USB y discos: cómo saber si es real

Lenovo once sold a USB flash drive advertised as having a fingerprint sensor for encryption. Subsequent disassembly, revealed through a Reddit discussion, showed that inside the drive was no more than a generic microSD card. There were no cryptographic chips, no AES engine, and nothing that could encrypt even a single byte. The fingerprint sensor only unlocked a Windows application; data read directly from the card appeared as plain text files.

This is not an isolated bug from a single manufacturer. It represents an entire market where the word "encryption" appears on packaging without any verifiable property behind it. This note explains what genuine hardware encryption looks like, the standards that certify it, and how to verify it in any USB or external drive before trusting sensitive data to it.

TL;DR Real hardware encryption requires a dedicated AES chip that encrypts data before it reaches flash memory. A fingerprint sensor that only unlocks a Windows app does not encrypt anything: the memory remains readable on any card reader. FIPS 140-2 or FIPS 140-3 validation in the NIST CMVP registry is the only verifiable proof of a real cryptographic module.

TCG Opal defines how a self-encrypting SSD protects its data key, keeping it out of the operating system. Commands like hdparm -I and sedutil-cli --scan can reveal if a disk supports and has active hardware encryption. What is hardware encryption? Hardware encryption is the technique where a dedicated chip inside a drive or USB flash memory encrypts and decrypts every byte with AES before writing or reading it, without the operating system or any program interfering or having direct access to the key.

Unlike software encryption (BitLocker in standard mode, VeraCrypt, LUKS), hardware encryption separates the encryption key from the main system. The key stays inside the controller, never exposing the operating system to the decryption key. In a truly secure drive, the key lives inside the controller, and even a memory dump of the device reveals nothing.

A fingerprint sensor alone does not define where the encryption happens. It can be the key for a legitimate cipher, as in a certified drive, or merely the trigger for a login screen application, as in the Lenovo case. The fingerprint identifies the person, but says nothing about what happens to the data afterwards. Why does it matter?

The risk isn't just losing a file. A company that buys 500 encrypted USB drives for its sales team and trusts the word is exposing customer data under a false sense of regulatory compliance. If the device is lost and never actually encrypted anything, the regulatory framework that exempts encrypted devices from breach notification no longer applies, and the company finds itself facing a breach notification that it thought it avoided.

The fingerprint sensor is not a security measure, but a decorative lock on a memory with no real protection, when the manufacturer never implemented genuine hardware encryption. This is not limited to cheap USB drives. In the past, academic research on commercial SSDs claiming hardware encryption found implementations that allowed bypassing the key without knowing the password, leading Microsoft to stop defaulting to hardware encryption reported by the disk for protecting BitLocker volumes.

The underlying problem is always the same: an unverified claim. The problem repeats itself outside of cheap USBs: external disks, portable SSDs, and even some home NASes use the word "encryption" on packaging without ever publishing a standard, verifiable algorithm, or certification number. The only way to know if there is anything real behind it is to ask about the standard, not the adjective.

How does it work inside? A self-encrypting drive (SED) separates two keys. The Media Encryption Key (MEK) is the one that actually encrypts data blocks with AES, almost always in XTS mode. That key is generated once, inside the chip, and never leaves it. The second key is the Key Encryption Key (KEK), derived from the PIN, password, or in well-designed biometric devices, from a secret the fingerprint sensor releases after verifying identity.

The KEK does not encrypt data, it encrypts the MEK, which is stored in a reserved area of the disk. Without the correct KEK, the MEK remains useless. This separation is what allows instant cryptographic wiping: wiping the MEK, without touching a single bit of the actual data, makes the entire content unreadable in milliseconds, because there is no key capable of decrypting it.

This is the mechanism behind secure erase functions in enterprise SSDs, and why a multi-terabyte disk can securely erase itself in seconds. The TCG Opal standard, published by the Trusted Computing Group, defines how a self-encrypting drive manages its key without exposing it to the operating system. The standard that defines how these pieces should behave in an SSD or HDD is TCG Opal.

USB flash drives do not necessarily follow Opal, designed originally for SATA and NVMe, but serious devices document an equivalent scheme: a controller with an AES engine, a KEK derived from the PIN or fingerprint, and a MEK that never crosses the USB bus in clear.