HAWK withdrawn after lattice attack

An AI-assisted lattice reduction forced the withdrawal of HAWK from NIST post-quantum signatures, while a Lean 4 kernel flaw allowed forged proofs of falsehood. Lesser reports covered a Perl entropy module fetching randomness over plain HTTP and a cryptography list revisit of Bitcoin mining concentration.

HAWK withdrawn from NIST PQC signatures

Anthropic researchers showed that HAWK-n key recovery reduces to the shortest vector problem in dimension n/2 + 1, halving the effective lattice dimension via an AI-assisted attack. The result prompted official withdrawal of the scheme from the NIST post-quantum cryptography signatures round. The pqc-forum thread ran to 18 messages from 11 participants.

Lean 4 kernel soundness bug allows forged proofs

A soundness flaw in the Lean 4 kernel lets metaprograms forge axiom-free proofs of False by abusing nested inductive projections, with a demonstration that 0 equals 1. The issue was disclosed on oss-security. Theorem-prover users must treat the kernel as untrusted until patched.

Perl Data::Entropy fetches remote entropy over HTTP

CVE-2026-18536 affects Data::Entropy versions before 0.010, which retrieve remote entropy sources over plain HTTP. An attacker can perform man-in-the-middle substitution of the entropy values. Perl developers relying on the module for randomness should upgrade or switch sources.

Bitcoin mining concentration and PQ migration

A cryptography mailing-list thread revisited Bitcoin mining concentration, noting roughly 62 percent controlled by three pools, together with the practical difficulties of migrating signatures to post-quantum algorithms. The exchange comprised 22 messages from 6 participants.