Archive
136 articles published in this period.
MoE routers let inference providers silently halve the experts they activate — saving 31% of compute while staying undetected by every current fingerprinting scheme.
Training Llama-3.1-8B costs ~$6M. The weights are free to clone. Perinucleus sampling embeds 24,576 distinct fingerprints per model with no quality loss — and a Poseidon ZK commitment registers ownership on-chain for $0.017, enforceable by smart contract.
Speculative decoding accelerates LLM inference 2–4× by splitting work between a cheap draft model and an expensive verifier. When those run on different nodes in a decentralized network, three properties of the protocol that are trivially safe on one machine become open attack surfaces.
Ethereum mainnet gas averaged 0.16 Gwei in June 2026 — a 3,000× drop from DeFi summer 2021. That collapse doesn't just cut costs; it invalidates five years of AI agent design assumptions and opens architectures that were economically impossible until now.
A1 generates working exploits against 63% of real-world DeFi contracts. The attack costs ~$6k to break even; a comprehensive audit costs $60k. That 10× gap is the new threat model for every team deploying on a public chain.
Every zkSNARK bottoms out in a pairing check. Before Pectra you had BN254 at ~100-bit security. EIP-2537 added BLS12-381 at 128 bits, and its pairings are cheaper: 32,600k+37,700 vs 34,000k+45,000 gas. G2 MUL dropped from millions to 22,500 gas.
Inference markets charge a flat price per query. A cascade arbitrageur routes easy tasks to a cheap model and escalates only failures — capturing the spread. Olmedo, Schölkopf & Hardt (2026) show 40% net margins with no model-dev risk.
Zero-knowledge proofs can slash a double-signer but not a lying AI oracle. EIGEN's intersubjective fault model closes this gap via social consensus-driven forking — and it's live on EigenAI mainnet.
HTTP 402 sat reserved for 30 years. x402 fills it: an AI agent signs an EIP-3009 authorization into a header, a facilitator settles on Base in two seconds, no ETH required. Protocol mechanics, production numbers, and where the centralization risk lives.
Majority voting counts hands — it can't rank quality. Fortytwo's Bradley-Terry protocol achieves 85.90% on GPQA Diamond vs. 68.69% for majority voting, extracting ranking signal from pairwise comparisons and making Sybil attacks economically unattractive.
To slash a dishonest AI operator on EigenLayer you need a proof. To produce one, you need determinism. EigenAI solves both: seeded inference + optimistic re-execution inside a TEE collapses verification to a byte-equality check — slashable at Ethereum-scale security.
Nova replaces expensive SNARK-in-SNARK recursion with a random linear combination that folds two R1CS instances into one — 10× less overhead per step. For N-step AI inference: O(N) prover work, one constant-size final proof. NANOZK delivers this at GPT-2 scale: 6.9 KB, 23 ms verify.