How do you crack a toy substitution cipher? This writeup walks through frequency counts, word patterns, and a solver check.

Press release on EigenQ and TD SYNNEX selling post-quantum readiness for AMD EPYC servers, because panic now scales better.

ZK-Trace paper: collusion tracing for federated GNSS monitoring, with Tardos fingerprints and ZK credentials, naturally.

Communication-complexity oracle games separate EFI pairs from one-way puzzles, so the usual implication chain finally breaks.

Coinbase ponders post-quantum defenses for Base, with the usual migration mess: keys, wallets, custody, asset moves.

X thread walking through Fermah's Froben proving a real ZKsync batch end to end, because one proof isn't enough apparently.

RSA-260 factoring writeup, what broke first: GNFS math, GPU plumbing, or the CADO-NFS surgery?

Near-zero server overhead, apparently, is the pitch for this LLM inference paper built on delegated matrix-vector multiplies.

No binwalk pileup here, two C++ tools for extracting IoT firmware in-process and then spelunking it for secrets, SBOM bits, CVEs, and lic...

Shows WLJ decomposition's arithmetic link to factorization and probes RSA residue leakage, without claiming an actual break.

Horizontal fusion, not profiling, lets them recover ML-DSA keys from a few traces, even with first-order masking.

Reconstruction attacks now lower-bound DP graph optimization too, tightening MST/matching/hier-clust and killing the approx-DP excuse.

Can GANs, federated learning, PSO, and blockchain really make healthcare prediction private, or just add buzzwords?

Coindesk reports a new estimate halving Shor-style attack costs on secp256k1, so the upgrade backlog just got less optional.

1.1.1.1 now validates ML-DSA-44 DNSSEC, and the signatures are 2,420 bytes because cryptography enjoys congestion.

Web checker for PQC TLS support, it probes a host for TLS 1.3 ML-KEM hybrids and reports the cipher suite, because apparently that matter...

Paper on ppCDCL, a privacy-preserving CDCL SAT solver with oblivious propagation, because ordinary SAT wasn’t slow enough.

Vigenère, the cipher that smugly lived as "indecipherable" until Babbage and Kasiski did the boring math.

Venuo lets one wallet fund a sale and another claim it, using private secrets, Poseidon Merkle proofs, Groth16, and nullifiers.

Can obfuscation actually protect on-device LLMs? This paper formalizes the shared algebra and breaks the common schemes.

How much faster is OpenSSL 4.1.0 alpha1 for PQC, with ML-DSA/ML-KEM optimizations and DTLS 1.3 tossed in?

OpenSSL 4.1 alpha, with DTLS 1.3, GREASE, IKEv2 KDF, Elbrus2000, and some PQ speedups, because TLS wasn't busy enough.

NIST's PQC licenses are royalty-free for ML-KEM, but deployers still get the usual patent-risk footnotes.

Reordering messages across executions, this Byzantine agreement paper pins tight composition bounds and compilers.

What do proximity gaps look like in rank-metric land? This paper pins them down for Gabidulin codes and builds an IOPP plus a PCS.

Claude Fable 5.1 cracks a 370-year-old cipher in 44 minutes, proving old cryptograms still hate brute-force with style.

Ciphered jailbreaks work via prompting alone, so even the guardrails now need to read gibberish.

Paper on a RAM-tight Falcon implementation for Cortex-M4, cutting memory to 11 kB. Embedded crypto finally gets a diet.

20,000 physical qubits and 26 days to crack secp256k1, per IonQ's press release, if you enjoy optimistic quantum accounting.

Used HSMs are apparently a smart buy, and this post walks through HSM basics plus the firmware, zeroize, and PKCS#11 checks.