Ethereum to deploy AI agent registration and verification standard 'ERC-8004' on mainnet this week

A smart contract specification standardizing the registration and trust verification of artificial intelligence (AI) agents on the Ethereum network is expected to be applied to the mainnet this week. As groundwork is laid to enable interactions among AI agents without a centralized intermediary, Ethereum’s role as AI infrastructure is expected to strengthen further.

According to Cointelegraph on the 27th (local time), Marco De Rossi, head of AI at MetaMask, said it is highly likely that the ERC-8004 standard could be deployed to the Ethereum mainnet as early as this week. Via a developer Telegram channel, he said, “ERC-8004 development has already entered a freeze phase, and it will be deployed to mainnet during the week, probably around 9 a.m. (U.S. Eastern Time) on Thursday.”

ERC-8004 is an Ethereum Improvement Proposal designed to allow AI agents to be registered and verified on-chain. With the standard in place, AI agents belonging to different organizations and platforms can discover one another, check reputations, and then interact even without a pre-established trust relationship. This process is implemented at the smart contract level without changes to the Ethereum network’s structure.

Ethereum’s official account also said on X on the day, “ERC-8004 enables AI agent discovery and portable reputation, supporting cross-organization interactions,” adding that “trust information can be propagated across the network.” Developers believe this could make an open AI agent economy a reality.

Existing protocols such as MCP (Model Context Protocol) and Agent2Agent supported communication and task execution among AI agents, but had the limitation of not addressing trust and discovery. ERC-8004 is designed to enable agent-to-agent collaboration even in trustless environments by standardizing these two elements.

Davide Crapis, head of AI at the Ethereum Foundation, said, “Ethereum is uniquely positioned to securely settle and guarantee AI-to-AI interactions,” adding that “trust and discovery mechanisms are key elements in building a cross-organization agent economy.”