A new preprint on arXiv studies the local leakage resilience of Shamir secret sharing. In this threat model, an adversary does not see entire shares but instead learns a bounded number of bits from each share. The paper sets this up for a scheme with (n) parties, a reconstruction threshold (k), and shares taken from a prime field (\mathbb{F}_p) where (p > n).
The abstract, as provided, cuts off before stating the main findings. It is therefore clear that the work targets a specific and practical attack scenario: partial information leakage from every share, rather than full compromise of a few shares. The parameter choices suggest the analysis applies to standard Shamir sharing over large prime fields.
Because the source is only the opening of the abstract, no concrete bounds or theorems can be reported here. Readers interested in the outcome will need to consult the full paper on arXiv.