MILP framework for picking better DFA fault bits and pushing attacks deeper, because brute force was getting too lazy.

Neural differential distinguishers, but evaluated under a ciphertext budget, with the usual multi-difference vs baseline comparisons.

Power analysis on constant-time MiMC over BN254, plus RNR vs ISW masking results that finally care about overhead.

Paper on SMT-aided quantum periodic distinguishers for block ciphers, with better round bounds. Classical cryptanalysis needed a hobby.

Can LLMs actually break ciphers? Schneier summarizes CryptanalysisBench and a few model-found attacks.