By combining artificial intelligence with holographic encryption, scientists have developed an extremely secure data protection system.
Their method encrypts laser beams into chaotic patterns, making them impossible to decode without a trained neural network. This innovation could revolutionize cryptography.
Holograms for next-level encryption
As the need for digital security grows, researchers have developed a new optical system that uses holograms to encrypt information, creating a level of encryption that traditional methods cannot penetrate. The advancement could pave the way for more secure communication channels that help protect sensitive data.
“From rapidly evolving digital currencies to governance, healthcare, communications and social networks, the need for robust protection systems against digital fraud is ever-increasing,” said research team leader Stelios Tzortzakis of the Institute of Electronic and Laser Structures, Hellas Research and Technology Foundation and the University of Crete, both in Greece. “Our new system achieves a special level of encryption by using a neural network to generate a decryption key that can only be generated by the owner of the encryption system.”
Revolutionary cryptography with artificial intelligence
In Optica, Optica Publishing Group’s journal of high-impact research, Tzortzakis and colleagues describe a new system that uses neural networks to extract complex encoded information stored as holograms. They show that a trained neural network can successfully decode complex spatial information in encoded images.
“Our research provides a solid foundation for many applications, especially cryptography and secure wireless optical communications, paving the way for next-generation telecommunications technologies,” said Tzortzakis. “The method we developed is highly reliable even under harsh and unpredictable conditions, addressing real-world challenges such as adverse weather that often limit the performance of optical systems in free space.”
Light interference for safety
The researchers developed the new system after discovering that when holograms are used to encode laser beams, the beam becomes completely scrambled and random, and the original shape of the beam cannot be recognized or retrieved by physical or computational analysis. They realized that this was an ideal way to securely encrypt information.
“The challenge is figuring out how to decode the information,” Tzortzakis said. “We came up with the idea of training neural networks to recognize extremely fine details in chaotic light patterns. By creating billions of complex connections, or synapses, in a neural network, we can recreate the original shape of the light beam. This means we have a way to generate decryption keys that are unique to each configuration of the encryption system.”
The researchers tested the new approach on handwritten digits and other shapes, such as stars. Credit: Stelios Tzortzakis, Institute of Electronic and Laser Structures, Hellas Research and Technology Foundation
High-power laser and ethanol magic
To create a physical system that can completely and chaotically mix light beams, the researchers used a high-power laser interacting with a small cuvette filled with ethanol. This liquid is not only inexpensive, but also produces the desired chaotic behavior over a short transmission range of just a few millimeters. In addition to changing the intensity of the light beam, light interacting with the liquid also exhibits thermal turbulence, which greatly enhances the turbulent interference.
Encryption and decryption were successful.
To demonstrate the new method, the researchers used it to encode and decode thousands of handwritten digits and other shapes, such as animals, tools, and everyday objects, from well-established databases used as references for evaluating image retrieval systems. After optimizing the neural network training and testing process, they showed that the neural network could correctly retrieve the encoded images 90-95% of the time. They say that this speed could be further improved by training the neural network more extensively.
The future of ultra-secure communication
The researchers plan to further develop the technology by adding additional layers of protection, such as two-factor authentication. Since the biggest hurdle to commercializing the system is the cost and size of the laser system, they are also exploring cost-effective alternatives to high-power, bulky, and expensive lasers.
Reference: “Optical information encoded in nonlinear chaotic systems discovered using neural networks” by Stelios Tzortzakis, Marios Manousidaki, and Panagiotis Konstantakis, February 19, 2025, Optica .
