Total security or giant files: how strong should a code be?

Imagine having to send a secret so precious that you want to protect it not only from today's thieves but also from the supercomputers of the future. This is the challenge that gave birth to KyFrog, a new encryption system designed to be almost unhackable. While current systems strive to be fast and lightweight, KyFrog focuses entirely on maximum security. To achieve this, it uses a mathematical method called 'Learning-with-Errors' (LWE), based on geometric problems that are extremely difficult to solve. KyFrog's trick is to use very large numbers and specific error distributions to make the task impossible even for future quantum computers. However, there is a small trade-off: to achieve this digital armor, the encrypted message becomes much heavier than usual. If standard codes are like a light postcard, KyFrog is like a massive steel safe. It is a bit slower to transport, but if someone tries to crack the safe, they would give up almost immediately. This experiment shows us that cybersecurity is a balancing act between speed and extreme protection.

KyFrog is based on the Learning-with-Errors (LWE) mathematical problem, a cornerstone of post-quantum cryptography. Unlike algorithms based on factorization (like current RSA), LWE is considered resistant to attacks using Shor's algorithm, which could destroy classical cryptography once quantum computers become powerful enough. KyFrog stands out due to its use of a lattice dimension (n = 1024) and a modulus (q = 1103) that create a massive security margin, estimated at around 2^325 for both classical and quantum attacks. This level of protection is significantly higher than current standards like ML-KEM (formerly known as Kyber). The main cost of this choice is the ciphertext size, which reaches about 0.5 MiB—an enormous value compared to the few bytes or kilobytes typically used. This makes KyFrog an ideal prototype for scenarios where absolute security is prioritized over available bandwidth.