This post is also available in:
עברית (Hebrew)
Protecting digital information usually depends on cryptographic keys, which are random sequences used to encrypt and decrypt data. The security of everything from smartphones and IoT sensors to industrial systems relies on keeping those keys hidden from attackers. Yet traditional hardware security solutions face a persistent problem: the keys are often stored in memory, making them potential targets for physical probing, reverse engineering, and other hardware-level attacks.
Researchers have now developed a hardware security architecture that tackles this problem differently. Instead of merely storing cryptographic keys, the system generates, conceals, and protects them using the underlying physics of the device itself.
The technology is built around spin-orbit torque (SOT) devices, a type of spintronic component that exploits both electrical charge and electron spin. When electrical current is applied, these devices can switch between magnetic states in a way that is inherently unpredictable. This randomness enables them to function as true random number generators, producing cryptographic keys that cannot be predicted in advance.
At the same time, manufacturing variations give each device unique physical characteristics. These variations act as a physically unclonable function (PUF), effectively creating a hardware fingerprint that is unique to each device. The result is a single platform capable of generating both dynamic and device-specific cryptographic material.
According to TechXplore, one of the most unusual aspects of the system is how it handles key access. When idle, the device does not store a readable key. To retrieve the correct key, a specific electrical excitation must be applied. This process changes the device’s internal magnetic state in an irreversible way.
In practical terms, any attempt to access the key leaves a permanent physical trace. An attacker trying to extract information would inevitably alter the device itself, making unauthorized access detectable after the fact.
From a cybersecurity and defense perspective, this approach is particularly relevant for connected devices deployed in exposed environments. IoT sensors, surveillance systems, communication nodes, portable authentication tokens, and military edge devices often face elevated risks of physical tampering. Embedding security directly into the hardware rather than relying solely on software protections could improve resilience against sophisticated attacks.
The researchers believe the same architecture could eventually support additional capabilities, including self-protection mechanisms and autonomous responses to tampering attempts. Their longer-term goal is to create security systems whose protection comes from the hardware’s physics rather than from layers of external defenses.
The research was published here.
