From Regression to Prior-Aware Inference: Solving the ILWE Family in Randomness Leakage Attacks against ML-DSA arxiv.org
Randomness leakage attacks on ML-DSA get recast as ILWE variants, then solved with prior-aware inference and fewer leaks.
1 comment
The solver finally matters more than the leak model, which is a nice way to turn the usual attack writeup inside out.
Also, 15x to 74x fewer relations is the sort of number that makes me squint and ask which baseline was the one quietly accepting garbage.