Crypto Implementation Vulnerabilities

A cryptanalysis paper on arXiv (cs.CR, 2026-10-05) shows that the CRT-based compartmented secret sharing scheme of Yang, Zhu, Fu and Xia (ISIT 2026), claimed to be ideal, is insecure. It exhibits three families of unauthorized subsets that reconstruct the secret, because the security analysis ignored some published polynomials. The only exception is a degenerate parameter choice where the whole participant set is the only authorized subset. Similar flaws appear in the same first author's hierarchical schemes (ISIT 2024, 2026).

An IACR ePrint paper (2026/2338, 2026-10-04) shows that "reconstruct-later" asynchronous verifiable information dispersal is insecure with the standardized Raptor codes R10 (RFC 5053) and RaptorQ (RFC 6330). For R10, every assignment of encoding symbols to honest parties contains a bad (f+1)-subset when K>⌊log2(2f+2)⌋. An implemented RaptorQ attack defeated an RFC 6330 decoder on every adversarial subset tested, and empirical attacks over F256 succeeded for every tested f from 9 to 45 with K=f+1. A preliminary Walrus design and an earlier Walrus testnet had used RaptorQ this way, and Reed-Solomon is suggested as a low-cost alternative.

An IACR ePrint paper (2026/2336, 2026-10-04) introduces query-limited RAM programs, a stateful generalization of quantum one-time programs, built from "one-shot programs" that unify one-shot signatures with the single effective query paradigm. It is a constructive theory paper, and the source describes no vulnerability, flaw or attack. It is likely a weak match for this stream.