AI breakthrough: Uruguay team reconstructs screen images via HDMI
Scientists from Uruguay have developed a technology that, thanks to artificial intelligence, allows for the recreation of images displayed on a computer screen. This advanced method analyses the electromagnetic radiation emitted by HDMI cables.
The privacy of computer users is extremely important, especially when it comes to protecting personal data and confidential information. Although access to this data often requires physical presence at the computer, the new technology may change this.
A team from the University of the Republic in Montevideo has developed an AI model that can capture an image from a computer screen by analysing electromagnetic radiation from an HDMI cable. They published detailed results of their research on the arXiv platform.
HDMI cable - to what extent can it be used for an "attack"?
Techniques based on electromagnetic radiation analysis are not new and have been used for analysing analogue signals. Now, with the use of artificial intelligence and cables, it has become possible to also analyse digital signals with 10-bit encoding, enabling more precise reconstruction of the image from the screen.
The team tested the new method using specialised software that allows reading text from a video signal, then comparing the obtained text with its original version. Although the error rate currently stands at about 30%, it is still a significant improvement over previous methods.
Tests were conducted using a system based on an Intel Core i7-10700F processor, 64 GB RAM, and an NVIDIA GeForce RTX 3090 graphics card with 24 GB VRAM. Capturing images with a resolution of 1600×900 pixels takes 15 seconds using only the processor and just half a second with the graphics card.
Who could be targeted by hackers?
What are the implications of these discoveries? According to scientists, the average user does not need to fear these types of attacks due to their low probability. However, organisations processing sensitive data should be aware of this risk and ensure additional security measures. The new technology may already be used in highly sensitive areas, such as industry or government agencies.
