Plain brute force cryptanalysis is destined to check on average half the possible secret keys that guard any cyber secret, including bitcoin fortunes. One could readily estimate how many microseconda checking one key will take, and hence use a key so large that checking half of the keys will last too long to be effective. This estimate underlies the security of all the ciphers used today, except the new class of pattern-devoid cryptography.
Brute force cryptanalysis estimates are based on the fact that a rejected key does not change the probabilties of any of the yet unchecked keys to be the right key. Alas, now comes AI and violates this premise. Each checked and rejected key changes the probability profile of the remaining keys, spotting keys of much higher probability to be correct. The better that AI capability the fewer keys that need to be checked in order to find the right one. While with ordinary brute force, the defender can readily estimate how long before the key is compromised, the defender of AI-accelerated brute force is without any clue how long before the key is compromised. When our clients internalize this threat they move towards the Pattern Devoid ciphers that are immunized against AI Accelerated Brute Force Attacks
תגובות