eBPF-based Linux crypto inventory, it uses randomness as a timing anchor to spot keygen and PQC ops at runtime.

Paper on purpose-limited secrets, tying them to constrained PRFs, FE, and payment tokens, because even secrets need policy.

Same PQC sermon, but framed as a CISO checklist, with the usual Y2K déjà vu and less hand-waving.

A real-world MPC workshop talk, because apparently multiparty computation now needs deployment war stories.

Prompt injection gets a fenced-off proxy and a closed command set, then a red-team SOC eval tries to break it.

ePrint on PQC OTAR over CCSDS links, showing endpoint processing time changes retransmission paths and timing.

Shows how EEZ + FCR compose into a fast, atomic cross-rollup payment path, because two waits were apparently one too many.

Paper on HE anomaly detection output policies, balancing threshold leakage, decision preservation, and CKKS stability. Fun times.

Paper on rare-event decorrelation for linear tests of low-complexity PRFs, finally giving AC0[2] bounds, because chaos needed math.

Can your hybrid crypto key or accept decision really depend on both PQ and classical parts? HyBind statically checks.

Less than a cent per signature, says this journal index, which mostly points to Medium-synced web3 and post-quantum essays.

Semi-quantum secret sharing that claims collusion security against a quantum third party and a dishonest classical player.

So NEAR shipped what kind of post-quantum asset, and does ML-DSA plus Intents actually buy anything?

Grinding challenges on a GPU, then piping FAEST v3 through CPU and GPU, because apparently signing needed a sidecar.

N2SF PQ migration framework classifies payload retention, rewrap, re-encrypt, comms, and signatures, with a toy implementation.

Bootstrapping gets fused into one GPU megakernel here, because apparently TFHE needed fewer global memory trips.

Paper on truncated-Montgomery PMNS reduction for modular arithmetic, claiming up to 15% AVX512 speedups. Naturally.

P-VMI shows adversarial images can keep steering multimodal LLMs long after removal, thanks to the KV cache doing what caches do.

Ethresear.ch post on prop-AMMs, from sub-slot freshness and maker priority to PBS-era MEV and more gatekept execution.

Puts PQ migration and classical vulns on one cost-of-delay ranking, so HNDL stops being a side spreadsheet.

MiniSHAKE defines Keccak-based hash/XOFs on Keccak-p[800,12], for when 1600 bits was apparently too much.

OpenSSH 10.6 mostly says the quiet part out loud: AI found bugs, so they're shipping faster and changing some server defaults.

Can you prove a site's AI training opt-in/opt-out without leaking the declaration? MachineWitness batches it into daily Merkle roots.

Sybils tighten the censorship-resistance bounds for multi-proposer protocols, redoing the fee-mechanism math in the process.

Computes weakest trust assumptions from minimal attacks and stopping sets, not just the usual assume-honest ritual.

16 Firefox extensions impersonated Rabby and OKX, then quietly exfiltrated wallet seeds and keys. Lovely marketplace hygiene.

Hand-coded AArch64 NEON for batched LSH, with a 4-way SPHINCS+ core that speeds signing without changing signatures.

BetweenCut does private heavy-node classification on trees with O(log log h) error, retiring the usual log h tax.

Can X-Wing fit on Cortex-M85, and what do the cycle counts, storage, and timing leaks actually look like?

Paper on LLM router metadata leaking sensitive topics, with measurements and mitigations, because even the headers gossip.