Researchers from the Korea Advanced Institute of Science and Technology (KAIST) have developed a device that controls access to encrypted data by time. The work was published in the journal Advanced Materials on September 10, 2026. The project leader is Professor Kyung Min Kim of the Department of Materials Science and Engineering, and the first author is graduate student Hwisu No.
The technology is based on an oxide transistor with dual programmable states (DUPOT). This is a single transistor in which two types of memory are set independently: long-term memory, which persists for a long time, and short-term memory, which changes over time. Conventional memory transistors store information by changing the threshold voltage - the voltage at which current begins to flow. The new device additionally controls the saturation current - the current level under fixed operating conditions. Therefore, it is possible to separately configure when the current will begin to flow and how strong it will be.
The two states are created by different mechanisms. With a positive voltage on the gate (the control electrode), electrons are trapped in a layer of hafnium oxide, which conducts current poorly. This shifts the threshold voltage and persists for a long time - this is how long-term memory works. With a negative voltage, oxygen ions move to the boundary of the insulating layer and the channel through which current flows, causing the saturation current to drop. After the voltage is removed, the ions gradually redistribute and the current recovers. This change is used as short-term memory.
On an array of several transistors, the researchers set how the current changes over time and established a readout criterion for each device. In an experiment with delayed-release encryption, the difference between the target signal and the background grew over time, and the letter "T" became readable after 8.4 seconds. In a time-bound variant, the correct information is available only within a set window: the false symbol "K" appeared after 1.5 seconds, and the target letter "T" after 37 seconds. The same array displays different data depending on the moment of readout.
The team also proposed a method for cloud environments, where data is stored on remote servers. In it, different time conditions are applied to the ciphertext (data transformed so that the original information cannot be recognized) in the cloud and to the decryption key held by the user. In the demonstration, the original information was recovered only when the ciphertext and the key matched in time condition.
The authors emphasize that the technology does not make it possible to revoke access to copies of information that have already been obtained. So far, the work demonstrates the fundamental possibility of temporal encryption at the level of an array of devices. The plans are to reduce the operating voltage, improve uniformity between devices and stability over repeated cycles, and then extend the approach to cloud and smart technology, as well as to neuromorphic devices that mimic the operation of brain neurons.