Ethereum is signaling a near-term rollout of ERC-8004, a draft standard that proposes a shared framework for AI agents to prove identity claims, accumulate reputation, and request validation using the network as a neutral reference layer.
In a Jan. 27 post on X, the blockchain’s official account said “ERC-8004 is going live on mainnet soon,” framing the standard as a way to enable agent discovery and portable credibility across platforms without centralized intermediaries.
ERC-8004 is not an Ethereum hard fork, and it does not change Ethereum’s consensus rules.
It is a standards-track ERC, meaning it defines application-layer interfaces and data formats that developers can implement in smart contracts and tooling.
That makes the phrase “going live on mainnet” easy to misunderstand.
In practice, a “mainnet launch” typically means:
On the ERC-8004 page, the proposal is labeled Draft and Standards Track: ERC, with a created date of 2025-08-13.
The author list includes Marco De Rossi and Davide Crapis, alongside contributors with Google and Coinbase email addresses shown in the document metadata.
The proposal also lists required standards: EIP-155 (chain identifiers), EIP-712 (typed signatures), EIP-721 (NFT-style identities), and EIP-1271 (smart contract signature verification).
Community signals suggest the rollout could happen as early as Thursday, Jan. 29, but confirmation will come from on-chain deployments.
The Ethereum Foundation did not immediately respond to a request for comment.
ERC-8004 aims to standardize a basic “trust layer” for agents, especially for scenarios where agents interact across different companies and ecosystems.
The proposal defines three lightweight on-chain registries: identity, reputation, and validation.
The identity component treats agents as ERC-721 identities, where each agent is represented by a token ID in a registry.
Metadata is pointed to via an “agentURI,” which can describe services, endpoints, and capabilities off-chain.
This design makes agents indexable in familiar Ethereum tooling. It also makes identity portable in a way that works across apps.
The reputation layer standardizes how clients can publish feedback about an agent and how others can query it.
That portability is the point: a reputation signal should be reusable across marketplaces and platforms.
The trade-off is predictable. If anyone can post feedback, then reputation becomes vulnerable to low-quality inputs and Sybil activity.
The proposal explicitly warns that consumers should filter reputation sources rather than treating raw signals as authoritative.
The third layer, validation, is meant for cases where “reputation” is not enough.
It defines a way for agents to request validation and receive responses from validators, which could represent many approaches, from crypto-economic checks to cryptographic attestations.
ERC-8004 does not mandate a single validation method. It standardizes the interface so different validation markets can plug in.
AI agents are moving from experiments to systems that touch real-world value: payments, data access, and automated decision-making.
Designers built most trust models for humans. They assume slow-moving identity, institutional accountability, and legible intent.
Agent-to-agent interaction breaks those assumptions. It scales faster, routes around organizational boundaries, and often operates through APIs and wallets rather than contracts and signatures on paper.
ERC-8004’s bet is that Ethereum can act as a neutral coordination layer for this shift.
It offers a shared format for identity and trust signals while leaving most compute and interaction off-chain.
Supporters have framed this as an expansion of Ethereum’s role beyond finance and toward infrastructure for interoperable AI services.
You’re All Set!
Thanks for signing up. We’ll be in touch soon with the latest insights.
