Instead of probing internals, this fingerprints LLMs from trivial one-token distributions and still tracks lineage through API chains.

Windows Insider builds ship SymCrypt SHA-3 and ML-KEM with Lean-checked Rust proofs, because proof debt is now product work.

Paper on Ethereum governance's hidden authority graph, with a deterministic reconstruction scheme, because visibility is hard.

Paper formalizes look-ahead bias as temporal non-interference for backtests and trading, with a decidable fragment, naturally.

Demo note showing the Q-Day ECDLP “break” still passes when IBM Quantum is replaced by /dev/urandom, so yeah.

Reuses one mechanism-design operator across markets, consensus, and LLMs, swapping admins for enforced invariants.

SecureCROWN does certified NN robustness in 2PC, hiding everything except the verdict, with Newton-Raphson and 200s runs.

An essay arguing multi-agent codegen dies on vague specs, and only verification or invariants save it, as usual.

Can you shrink UBQC brickwork under the blinding layer without leaking anything? BPBO says yes, via certified local resynthesis.

Apparently hardware benchmarks need Merkle receipts, every number is tied to an append-only evidence graph.

ISCA 2026 Cryptolets tutorial page, for the open hardware answer to one-off crypto accelerators: chiplets, flows, and benchmarks.

Paper on deterministically reconstructing DAO governance outcomes from public Ethereum evidence, turning reconstruction into a verificati...