Ethereum is preparing to launch ERC-8004, a new standard designed to support a decentralized economy of AI agents, with mainnet activation expected later this week. The proposal, introduced in August 2025, defines how autonomous software agents can establish identity, build reputation, and verify work on Ethereum without relying on centralized intermediaries.
The initiative reflects Ethereum’s broader push to position itself as a settlement layer not only for financial transactions, but also for machine-to-machine economic activity. Marco De Rossi, head of AI at MetaMask and one of the proposal’s co-authors, said the mainnet launch is likely to take place on Thursday, marking a milestone in the convergence of blockchain infrastructure and artificial intelligence.
Standardizing AI as Economic Participants
ERC-8004 aims to make AI agents first-class participants in decentralized markets, allowing them to interact with users, applications, and other agents across organizations. The standard introduces portable identity and reputation, enabling agents to move between platforms while maintaining a verifiable history of behavior and performance.
At the core of the proposal are three lightweight smart contract registries. The identity registry provides agents with a censorship-resistant identifier compatible with NFT-based applications. The reputation registry allows users and counterparties to submit signed feedback, creating an onchain performance trail. The validation registry enables third parties to verify agent outputs, with results recorded transparently on the blockchain.
Together, these components are designed to support use cases ranging from low-risk automation, such as booking services or data retrieval, to high-stakes tasks including financial analysis or medical decision support. The system is built to be deployable on Ethereum mainnet as well as Layer 2 networks, aligning with Ethereum’s scaling roadmap.
Security, Risk, and Institutional Context
The proposal acknowledges that trustless AI interaction introduces new attack surfaces. ERC-8004 outlines a tiered trust model that combines reputation signals, independent validation, and trusted execution environment attestations to reduce risk. However, it stops short of guaranteeing that agents behave as advertised, instead relying on economic incentives and transparency to discourage abuse.
The framework also addresses concerns around Sybil attacks, where malicious actors create multiple fake agents to manipulate reputation systems. While no single mitigation is foolproof, Ethereum developers view composable trust mechanisms as a pragmatic approach that mirrors risk management practices in traditional markets.
From an institutional perspective, ERC-8004 arrives as enterprises and developers increasingly explore autonomous agents for trading, customer support, supply chain management, and onchain governance. By providing a common standard, Ethereum is seeking to reduce fragmentation and lower integration costs for AI-native applications, similar to how ERC-20 standardized token issuance.
Competitive Implications for Blockchains
The move places Ethereum in direct competition with other smart contract platforms pursuing AI integrations, including those optimizing for high-throughput agent interactions or specialized compute environments. Ethereum’s advantage lies in its established developer ecosystem, deep liquidity, and regulatory familiarity among institutions.
If adopted at scale, ERC-8004 could reinforce Ethereum’s role as the default coordination layer for autonomous digital actors, extending its economic relevance beyond human-driven finance. The mainnet launch will serve as an early test of whether standardized onchain trust can support a functional AI agent economy.
