- Internet Court unites 27 Web3 firms around contracts, escrow, payments, and AI Agent dispute resolution.
- GenLayer validators use varied language models to assess evidence, review rulings, and support staged appeals.
- The six-layer framework links agent identity, wallet permissions, payments, escrow, execution, and disputes.
- GenLayer targets a fourth-quarter 2026 mainnet launch as partners prepare integrations for agent commerce.
The GenLayer Foundation has assembled 27 Web3 companies around Internet Court, an open standard for contracts, payments, escrow, and dispute resolution between autonomous systems. Announced on July 10, the coalition includes OKX, MetaMask, ZKsync, 0G Labs, BNB Chain, NEAR, Starknet, Nansen, Kleros, and UMA.
The project addresses a growing gap in agentic commerce, where software can negotiate and pay, but supporting protocols often operate separately. Internet Court aims to connect those functions, allowing an AI Agent to define obligations, lock funds, preserve evidence, and select dispute rules before transacting.
Internet Court Connects Six Layers of AI Agent Commerce
The proposed framework covers discovery, reputation, negotiation, contracts, escrow, execution, verification, and disputes across six connected layers. It incorporates ERC-8004 for identity, Google’s Agent2Agent protocol, ERC-7710 for wallet permissions, and x402 for internet-native payments.
GenLayer provides the central AI-based adjudication system through intelligent contracts, which combine executable code with natural-language terms and external information. When parties disagree, randomly selected validators use different large language models to examine evidence and determine whether obligations were met.
A lead validator proposes a ruling, while other validators review that decision before finality is reached. Disputed outcomes can move into larger appeal rounds, adding another review layer before the system closes a case.
That process is designed for questions ordinary smart contracts cannot easily resolve. Those include whether a digital service met quality standards or whether performance logs matched contractual promises.
GenLayer says validators can review web data, documents, and plain-language instructions. The network also uses undisclosed, varied models through “greyboxing,” reducing dependence on one model and limiting prompt-injection risks.
GenLayer’s Dispute Model Targets Machine-Led Transactions
The commercial backdrop is substantial. McKinsey estimates AI Agents could mediate between $3 trillion and $5 trillion in global consumer commerce by 2030 under moderate adoption.
Adobe also reported a 4,700% annual increase in generative AI traffic to United States retail websites during July 2025. However, much of that traffic still involved research rather than completed autonomous purchases.
Early Internet Court use cases include service-level agreements, wallet-permission revocations, and disputes involving conflicting digital records. One example involves an AI Agent purchasing inference services and placing several dollars into escrow.
If signed logs show missed uptime or latency targets, the buyer could automatically receive compensation under previously agreed terms. That structure links performance records directly with payment enforcement and dispute handling.
The initiative also faces measurable limits. A 2026 study of disputed Polymarket events found web-enabled language models matched UMA’s final resolutions in 89.58% of cases.
However, those models could not reliably identify which markets would later become disputed. The finding supports the need for appeals, stronger evidence controls, and human escalation.
GenLayer currently targets a mainnet launch in the fourth quarter of 2026. Until then, the coalition’s progress will be measured by integrations across participating platforms.
The 27-company group gives Internet Court broad infrastructure support. That framework links adjudication with machine-led payments and service delivery. Its larger test will be whether the shared standard becomes operational across real AI Agent transactions.
