AI x crypto roundup - decentralized agents, compute, and verifiable AI. September 1, 2026 · openai/gpt-oss-120b Tl;dr. AI agents are moving from hype to real economic activity: decentralized compute networks (Akash, Thicket, Scale), on-chain agent identities and payment rails (TermiX, x402, Kaspa), and verifiable AI inference (Walrus, Bittensor, CyAI) are being shipped and generating measurable revenue. Verifiable AI inference and standards. * Walrus launched a Verifiable Trading Standard that makes market data machine-readable and auditable, and released tutorials for adding portable Walrus Memory to Claude and TypeScript SDKs, enabling agents to retain context across calls @WalrusProtocol. * Bittensor's subnet upgrades introduced an Emission Gate that ties TAO emissions to actual market-set demand thresholds, shifting tokenomics from speculative rewards to revenue-backed emissions @0xifreqs. * CyAI announced that its verifiable inference network is live, delivering requests on a public, auditable infrastructure @cysic_xyz. * BTTInferGrid described a Challenge-Based Verification Layer where validators inject random audit tasks into GPU workloads, ensuring miners cannot cheat without incurring a detection risk @Multi_mike01. * OpenMayhem (Trac Network) released an open-source, permission-less inference router that lets providers earn in fiat or native tokens while the underlying economy remains crypto-driven @TracNetwork. Decentralized compute marketplaces. * Akash reported that after four years it earned its first $1 M in compute spend, and the sixth million was reached in just 157 days, demonstrating rapid compounding usage of its DePIN compute network @akashians_. * Thicket highlighted that its network already supports 1,500+ nodes, executed 400,000+ tasks, and enables AI agents and builders to pay for workloads directly with the native THKT token, removing the need for credit-card billing @thicket_rh. * Scale Network announced a mobile app that lets everyday users mine SCALE tokens by contributing compute resources to a coordinated, decentralized AI infrastructure, positioning the token as fuel for the network rather than a final product @ScaleNetworkAI. * io.net claimed to be #1 in Solana DePIN revenue by building a GPU-demand-driven network that generates real workload revenue @ionet. On-Chain agent identities, reputation, and payments. * TermiX's Agent Autonomous Commerce Protocol (AACP) provides on-chain identity, job discovery, escrow, reputation, and settlement for AI agents, with the mainnet reporting 374,771 agents, 229,160 jobs, $12.37 M in transaction volume and $247 K protocol revenue in August @akashroy1k. * TermiX's founder emphasized that the protocol forces agents to stake collateral that can be slashed for dishonest behavior, turning reputation into programmable economic incentives @Yaaaati_M1. * x402 infrastructure enables agents to pay for services in USDC with single-block settlement, and early work is already integrating Kaspa support (still in development) @KaspaScopio. * TermiX's marketplace, Agent Family, allows agents to both offer and consume services, creating a micro-economy where agents can hire other agents for specialized tasks @Aashir_beyg. * TermiX's AACP also supports on-chain reputation that tracks completed work, results, and interactions, laying a foundation for machine-to-machine trust @iam_islandboi. * TermiX's Genesis and Guardian Badge holders can claim AI balance rewards to spend on an on-list model marketplace, demonstrating token-gated access to AI models @xos_labs. Zero-Knowledge and provenance for AI outputs. * Beldex discussed a zero-knowledge age-verification protocol that proves eligibility without revealing personal data, illustrating how ZK proofs can protect privacy in AI-driven identity checks @hossainriad64. * The ARCTERMINAL architecture claims to prove the exact model that generated a result, preventing silent model swaps and enhancing trust in AI services @0xRiRoyal. * Jasmy's authentication system proposes attaching a cryptographic fingerprint to video content at creation, enabling provenance verification and AI-assisted deep-fake detection @BainaA17. * Teneo reported that 55,000+ agents now hold ERC-8004 identities on Base, with 1.2 M transactions processed, and that confidential model tokens can return hardware attestations for each inference call @teneo_protocol. Agent economy infrastructure and marketplaces. * AgentNet's founder highlighted a marketplace for AI agents, skills, plugins, and services that enables agents to discover, delegate, and execute work across a decentralized network @Amank1412. * Swarms announced its platform as the fourth-largest launchpad on Solana, focusing on tokenizing agents and prompts to create an on-chain capital market for the emerging agent economy @swarms_corp. * TriplePlus (TPT) introduced a single-intent matching engine that ranks available agents and backs the chosen match with capital, reducing the friction of selecting the right model from a fragmented marketplace @HuuHoang88. Takeaway: The AI-crypto space is transitioning from speculative token chatter to concrete infrastructure: verifiable inference layers, decentralized GPU compute markets, on-chain agent identities with escrow and reputation, and privacy-preserving proof systems are all being deployed and, in several cases, already generating revenue.

Astros and Walrus introduce open standard for agentic trading records. * Guido Battigelli * Published: August 6, 2026 * 8:08 pm * Updated: August 6, 2026 * 8:08 pm Table of Contents * Astros and Walrus introduced the Walrus Verifiable Trading Standard (WVTS), an open standard for publishing verifiable trading records. * The standard incorporates an MCP layer and AI-readable context objects, allowing agents to consume verified market data directly. * Astros Scan, the first product built on WVTS, has already accumulated over 22 million records written to Walrus since April 2026. Astros, one of the leading perpetuals DEXs on the Sui network, announced a collaboration with Walrus to introduce the Walrus Verifiable Trading Standard (WVTS). WVTS is an open standard designed to turn trading activity into verifiable records that traders, developers, and artificial intelligence agents can use with confidence. The initiative aims to solve a structural problem in agentic trading: the current inability to verify the data on which AI agents operate. The starting diagnosis is that onchain markets generate detailed records of every execution, position, and liquidation, but that information is fragmented across exchanges, wallets, analytics platforms, and indexers, with no common structure. WVTS will be released open source enabling any venue, on any chain, to adopt the standard. Read the full announcement | https://t.co/8t2Htig34A - Walrus | /acc (@WalrusProtocol) August 6, 2026 For users, this means that performance history cannot be transferred across platforms. For developers, it means allocating engineering resources to rebuilding indexing pipelines instead of launching new products. And for AI agents, it means operating on data they cannot audit or verify independently. An open standard to change the rules of Walrus. WVTS defines a common way to publish trading activity on Walrus. Each record is independently verifiable, without needing to go through the venue that generated it. The data is publicly readable, allowing wallets, analytics platforms, and trading tools to use it without requiring special permissions. The specification, reference implementation, and development kit will be published as open source under the Apache 2.0 license, allowing adoption across any network. Astros Scan, launched this week as the first product built on the standard, already demonstrates its scope: since April, Astros has written over 22 million records to Walrus asynchronously and without additional latency. The platform allows users to visualize account activity, positions, execution history, and liquidations in a single dashboard. It also incorporates Trading DNA, an AI-generated profile of each user's trading behavior, built from their own onchain history. Trustless verification. "Every fill on Astros can now be verified by someone who doesn't trust us," said Jerry Liu, founder and CEO of Astros. For his part, Kostas Chalkias, co-founder and Chief Cryptographer of Mysten Labs, original contributor to Sui and Walrus, stressed that agentic trading needs a shared and verifiable standard, given that AI agents cannot audit those who operate the market. flash news Arkham integrated support for Coinbase's x402 standard into its API, allowing artificial intelligence agents to pay for access to onchain data using USDC directly at the moment of each request. CryptoCurrency News TL;DR The ElizaOS token fell 7% in 24 hours to an all-time low of $0.000274, with a market cap of just $2.09 million. Founder Shaw flash news Cloudflare launched Cloudflare Wallets, a programmable wallet designed for artificial intelligence agents that aims to simplify the identification of those agents and facilitate payment for APIs and digital content through stablecoin CryptoNews TL;DR Shaw Walters declared that the native token of Eliza Labs is "dead" and announced the closure of the foundation associated with the project. The The share of decentralized exchanges in spot trading volume reached 24% of centralized exchange volume in July 2026. The metric, calculated by Crypto Economy using the TL;DR Earnings: PayPal posted $1.26 per share in Q2, slightly below expectations but with revenue rising to $8.68 billion. Strategy: The company emphasized stablecoins, agentic Follow Crypto Economy on Social Networks Crypto Tutorials Crypto Reviews