62 stories
Show: Summaries / Headlines only
EMVCo's Security Working Group puts a code-breaking quantum computer at least two decades away and sets no migration date for chip cards. The authors built their timeline on three qubit counts, one of which IBM gives in logical qubits.
StarkWare spent 10,000 satoshis from a lock built on hash-based signatures, in a Bitcoin transaction mined by MARA Pool. A flaw in the original script, which left one of two signature rounds unenforced, was fixed on October 7.
Researchers have published flaws in new post-quantum code since 2024. For long-term keys, a classical algorithm beside the new one limits the damage, at the cost of extra data, code, and a second migration. The NSA does not require hybrids.
wolfSSL's optimized ML-KEM code for two processor families compared only part of each incoming ciphertext. In tests on a reused key, Bhabani Sankar Das recovered 98.0% to 98.5% of its secret coefficients. Version 5.9.2 fixes both flaws.
Poseidon was designed to be cheap inside zero-knowledge proofs. Justin Drake attributed the switch to faster proof systems for standard hashes. The hash-based design that used Poseidon has not been deployed on Ethereum.
Google marked post-quantum key exchange as done on its API endpoints and internal service traffic. It set certificate, identity, and hardware milestones through 2028 and said it projects most services will meet their domain dates.
RFC 10024 marks only X25519MLKEM768 of its three hybrid groups as recommended and marks two pre-standard Kyber code points as discouraged. Server authentication, which uses certificate signatures, is unchanged.
Weiqiang Wen and Jinwei Zheng proved, for quantum computers, an equivalence between Module-LWE over the ring family ML-KEM uses and a new structured problem. They presented no algorithm that solves either problem and no attack.
The Amazon Web Services researcher labeled his paper a preliminary draft, marked four proofs as sketches, and cited personal communications for its lattice corollary. He described no attack on ML-KEM, ML-DSA, or any NIST parameter set.
The insurer's researchers put the code-breaking machine at roughly 1,250–1,450 logical qubits, against 4,700–7,500 for leading financial applications. The Google researchers they listed as a source reported fewer than 1,200 at one trade-off point.
