Paving the way for "AI Agent Economy": Ethereum's new standard "ERC-8004" is about to go live

Ethereum announces that a new technical standard ERC-8004, specifically designed for “Trustless AI Agents,” will be officially deployed on the mainnet this week, marking a key step forward in the deep integration of AI and blockchain. ERC-8004 was first proposed in August 2025, with the core goal of enabling AI agents to interact and collaborate across platforms and organizations within the Ethereum ecosystem, thereby shaping a permissionless, decentralized AI economy where AI is no longer just a tool but an “agent” capable of participating independently in economic activities.

Ethereum official stated on social platform X:

By granting AI agents discoverable identities and portable reputation mechanisms, ERC-8004 allows AI agents to freely interact across organizations and ensures their credibility can be verified everywhere. This will unlock a global AI service market without gatekeepers.

ERC-8004 is going live on mainnet soon.

By enabling discovery and portable reputation, ERC-8004 allows AI agents to interact across organizations ensuring credibility travels everywhere.

This unlocks a global market where AI services can interoperate without gatekeepers.

— Ethereum (@ethereum) January 27, 2026

The design of ERC-8004 does not assume all AI tasks carry the same risk but introduces a pluggable, tiered trust model that adopts appropriate verification and trust mechanisms based on task risk and security requirements. In other words, low-risk tasks (such as automatic pizza ordering or information queries) and high-risk scenarios (such as medical diagnosis or financial decision-making) can apply different levels of trust and verification, avoiding systemic risks associated with a one-size-fits-all approach. Davide Crapis, head of AI at the Ethereum Foundation, pointed out that Ethereum has a “unique advantage” and the ability to become the underlying platform for interaction, settlement, and trust among AI entities. He stated that connecting the AI community with decentralized infrastructure has always been part of Ethereum’s long-term mission. ERC-8004 to Launch as Early as Thursday Although Ethereum did not specify an exact timeline in the announcement, Marco De Rossi, head of MetaMask AI and one of the co-authors of ERC-8004, revealed that the mainnet deployment is “most likely scheduled for Thursday at 9 a.m. Eastern Time (10 p.m. Taipei Time).” The logic behind ERC-8004 is simple yet profound: if AI agents are to independently conduct transactions, coordinate resources, and execute tasks in the future, they must have long-term identities similar to real humans, wallets, or smart contracts, along with a recognized credit evaluation system. Currently, many enterprises deploy AI agents internally, but these systems mostly rely on closed identity lists and API keys for integration. While effective within a single company, trust mechanisms tend to break down when involving cross-supplier, cross-chain, or even cross-border collaboration. Breaking the “Communication Standard” of Closed Systems To address this pain point, the core design of ERC-8004 includes three lightweight registries that can be deployed on the mainnet or any Layer 2, serving as mechanisms to discover and trust AI agents in untrusted environments: Identity Registry: Utilizing ERC-721 (NFT), assigns each AI agent a unique on-chain identifier. This “digital passport” records the AI agent’s functions, access methods, and supported protocols, with anti-censorship features, and the ownership of the identifier can be transferred or delegated, allowing AI identities to migrate across platforms. Reputation Registry: Both humans and other machine clients can leave performance evaluations for AI agents here. These raw signals are stored on-chain for free access by various applications. The purpose is not to directly rank AI agents but to make “credit data” transparent and reusable. Verification Registry: A mechanism for AI to apply for “proof of capability.” Verifiers can be staking service providers, machine learning proofs (ML Proofs), or other verification systems. Verification results are recorded on-chain, allowing users to see what has been verified and by whom.