“What if the real unlock for AI x crypto is not speculation — but proving intelligence can be trusted?” @sachitakamura sits down with @dcbuilder for a deep dive into the intersection of AI and crypto, from DC’s work building the ZKML community to why zero-knowledge machine learning could become a core trust layer for AI on the internet. They break down how ZKML makes it possible to prove that a machine learning model produced a specific output from specific inputs — without forcing everyone to rerun the full computation themselves. The conversation also explores why this matters for model accountability, transparency, and verifiable AI at scale, plus the projects pushing the space forward, from Modulus Labs and Giza to EZKL and new research around proof of inference.

Checking an AI answer is never free, it just charges you in a different currency each way. a) Heavy cryptographic proof (zkML) charges compute. b) Trusted-hardware approaches (TEEs) charge a dependency on the chip maker. c) Consensus-scoring charges accuracy, because it cannot actually prove the model ran. Ours charges a small overhead per response, a company claim we have not had independently benchmarked, running on testnet today. Every one of these is a real cost, and the only wrong move is pretending your option is free.

Excited to share that @icme_labs has been selected for the @0G_labs Apollo Accelerator, backed by @theBBFund and veterans from Stanford's blockchain community @StanfordSBA. 🔥Only 10 teams selected out of hundreds🔥 #ZKML# We're building PreFlight: cryptographic guardrails for AI agents. Plain English policies compiled to formal logic used for enforcement; with a succinct ZK proof of every decision. LLM judges don't work. Humans-in-the-loop, no way! Agents that handle anything important need cryptographic guardrails. That's PreFlight. #0gApolloAccelerator#

Prediction: You won’t lose your job to AI, until your company starts using PreFlight (Automated Reasoning + ZKML). Here is why. Imagine the company fires the HR person and replaces them with an agent. Over time that agent could, and would likely start hallucinating. How would management know that its decision were based on facts (company HR policy)? “AI is a black box!” With AR, they could turn that policy into formal logic and make sure the agents outputs correspond to facts, precisely. No more guessing if it hallucinated. But who wants to go in and check a thousand AI outputs against formal logic proofs - to audit all of this? “No one.” PreFlight wraps these formal proofs into one succinct proof. With ZK it can be verified in under 1s. “All of our policy rules have always been followed.. by the new HR agent.. shi..” Personally, I am unsure if we can undo this invention. The moment you had agents that are guaranteed to follow facts and succinctly verifiable to have done so, is the moment you have reliable AI coworkers..

Decentralized AI should be invisible users should send a request and receive the result at a clear price, latency, and service level everything else should happen in the backend this is why i’m watching projects that hide decentralized infrastructure behind familiar products: - @ambient_xyz routes @OpenRouter requests across independent miners, checks outputs, penalizes failures, and reruns jobs when necessary. At one point, it processed more than 20B Kimi K2.7 Code tokens per day - @engyai offers an OpenAI-compatible gateway backed by distributed compute. Its @Alibaba_Qwen traffic has exceeded 20B tokens on several days - @OpenGradient lets users choose between signed execution, TEE attestations, and ZKML without changing the basic inference experience - @NEARProtocol AI uses hardware attestations to verify that an approved model ran inside a confidential environment - @darkbloomai routes encrypted requests to independently operated and hardware-verified machines i also see early progress beyond inference: - @Pluralis distributed RL rollouts across 14 Macs in 4 countries while one B200 updated the model - @opentensor supports subnets with various usecases for modern AI infra - @Pareton_ai rewards inference-engine improvements only when they reduce GPU time without weakening quality or breaking the SLA a decentralized network only works commercially if lower compute costs outweigh coordination overhead, failed attempts, and weaker service quality most enterprises will not send sensitive info to unknown operators for a small cost reduction blockchain only earns a role if it improves open access, reputation, collateral, verification, or settlement a token only earns value if the network actually needs it for fees, collateral, settlement, or ownership of the economics i think decentralized AI will not replace centralized cloud infra a hybrid model looks more realistic, customers use an accountable service while portable workloads are routed across independent suppliers behind the interface

This week’s episode features DC Builder (@dcbuilder), Research Engineer at the Worldcoin Network (@worldnetwork). We dive into Worldcoin’s mission to democratize digital identity and finance worldwide, and why proving personhood could become one of the most important primitives for the next era of the internet. The conversation explores the challenges behind verifying real humans at global scale — from web of trust systems to biometrics — and how Worldcoin approaches identity, privacy, and fair wealth distribution. DC also breaks down the role of Semaphore, zero-knowledge proofs, and privacy-preserving infrastructure in making digital identity usable without turning it into surveillance. We also dig into the emerging ZKML space, the intersection of crypto and AI, and why machine learning, verification, and decentralized systems may become one of the most exciting frontiers in Web3. A deep episode on identity, trust, ZK, AI, and the infrastructure needed to make the digital world more human.

Great seeing @zk_albi from the @icme_labs team leading two talks. One as a guest speaker on new cutting edge efficient recursion for JOLT. The other his own work on ABBA (lattices). These talks are on YouTube: search “zkSummit Rome”. In our Jolt Atlas (zkML JOLT) repo - upstream work from the a16 helps us greatly 🙏 this year we have ZK (privacy) coming from folding scheme (cross stream), efficient recursion, and a Jolt prover that can work on constrained devices. Down stream: a lot of the core focus has been on EC primitives. A lot of the work needs porting to lattice. One example: add privacy (ZK) in the lattice setting. Another is succinct proofs. Getting one GPU proving one cpu core in realtime.. is likely happening these year.. Let’s go! 🔥🔥🔥

Listened to the most recent @zeroknowledgefm "Is ZK dead? Or has it just begun?" Deep research takes time. The funders were early to the vintage. Now that vintage is becoming real world products. Will make @tarunchitra bullish on ZK again 😆 and ZKML. 'Automated reasoning' turns natural language into formal logic and can be used as AI guardrails that can't be bypassed or ignored. Will CC when our paper drops mid-month🤞.