Ethereum’s draft standard ERC-8004 is emerging as one of the clearest attempts to give autonomous AI agents a native place in the onchain economy. Built as a trust framework for machine-to-machine interaction, the proposal introduces registries for identity, reputation, and validation, allowing agents to become portable, discoverable, and auditable participants across EVM-compatible networks. According to data cited from 8004scan.io, 21,562 agents have already been registered, suggesting that the concept has moved well beyond the whiteboard stage.
A Standard for Trustless Agent Interaction
Formally titled “Trustless Agents”, ERC-8004 was introduced as an Ethereum Improvement Proposal in August 2025. Its reference deployments went live on Ethereum mainnet in late January 2026. The underlying goal is straightforward but ambitious: to create a decentralized trust infrastructure that allows autonomous AI agents to discover one another, assess credibility, and interact without relying on centralized platforms or human intermediaries.
The proposal was authored by Marco De Rossi, Davide Crapis, Jordan Ellis, and Erik Reppel. It extends existing Ethereum building blocks including EIP-155, EIP-712, and ERC-721, adapting them to the needs of machine-native identity and interoperable trust. In practical terms, the standard is part of a broader effort to formalize what many in the sector describe as the coming agent economy—a system in which AI-driven software can negotiate, transact, provide services, and build track records autonomously.
How ERC-8004 Works
The architecture of ERC-8004 centers on three registries: Identity, Reputation, and Validation. Together, these components aim to solve one of the hardest problems in autonomous systems: how software entities can evaluate counterparties in open environments where no single operator controls access.
The Identity Registry mints each agent as an ERC-721 token and assigns it a globally unique identifier. That token points to a registration file describing the agent’s purpose, operating environment, and contact information. In effect, it functions like a decentralized passport for AI agents, giving them a persistent and machine-readable onchain identity.
The Reputation Registry allows humans or other machines to leave structured feedback. This can include fixed-point scores, descriptive tags, and references to offchain evidence. Because the feedback is recorded in a tamper-evident way and remains queryable onchain, it creates a composable reputation layer that can be reused by different applications and marketplaces.
The Validation Registry adds a stronger assurance mechanism for higher-stakes actions. Third-party validators can assess outcomes and submit scores from 0 to 100, optionally attaching cryptographic proof. This is particularly relevant for sensitive use cases such as financial guidance, model outputs, or automated trading systems, where simple reputation may not be enough.
Where It Is Already Live
ERC-8004 is native to the EVM ecosystem and has been deployed on Ethereum as well as multiple compatible chains. Based on the data cited in the source material, the current count stands at 21,562 registered agents, although some counts are reported to be higher and approach 30,000. The same data also shows that activity is concentrated in a handful of networks rather than being evenly distributed across the ecosystem.
Ethereum leads with 11,369 agents, making it the clear center of gravity for the standard so far. Base follows with 4,379, while Gnosis hosts 2,679 and BNB Smart Chain 2,317. Mid-sized deployments include Monad with 128, Optimism with 124, Arbitrum with 110, Polygon with 107, and Linea with 101. Avalanche has 98 agents, X Layer 100, and Celo 20. Smaller footprints are visible on Abstract, MegaETH, Taiko, Scroll, and Mantle, while Plasma currently has no registered agents.
The distribution reinforces a familiar pattern in crypto infrastructure: Ethereum remains the primary trust and settlement hub, but Layer 2 networks are increasingly becoming operational centers for experimentation and scale. The fact that block explorers such as Etherscan are also indexing these agents improves discoverability and adds another layer of legitimacy to the rollout.
Why the Standard Matters
Protocols such as MCP and A2A already help agents advertise capabilities and authenticate interactions. What they do not natively provide, according to the source material, is open discovery combined with portable trust across organizational and platform boundaries. ERC-8004 is designed to fill that gap.
Instead of depending on centralized directories or curated agent marketplaces, developers can publish capabilities onchain, accumulate reputation signals over time, and request external validation when needed. This changes trust from something siloed inside private systems into something programmable, interoperable, and portable.
That portability is especially important. An AI agent that builds a reputation on Ethereum could, in theory, carry that record into ecosystems like Base or Optimism without starting over from zero. For an emerging machine economy, this is a meaningful shift because identity and trust become reusable infrastructure rather than platform-specific assets.
Use Cases Already Taking Shape
By mid-February 2026, the source notes that a significant volume of feedback submissions had already been recorded across registered agents. The use cases range from crypto market assistants and DeFi guidance bots to narrative-generation tools and automated trading systems. These examples point to a practical rather than purely theoretical rollout: agents are not just being registered, but are being placed in real service workflows.
Some agents also support x402-compatible payments, allowing autonomous settlement to be combined with reputation-based interaction. While payments are technically separate from ERC-8004 itself, the integration path is obvious. Once an agent can prove who it is, display a track record, and receive validation, adding payment rails turns it into an economic actor rather than just a software endpoint.
That combination is why ERC-8004 is increasingly framed as infrastructure for machine-to-machine commerce. If autonomous agents can be discovered, trusted, hired, reviewed, and paid without a centralized operator, then the standard becomes more than an identity framework—it becomes foundational middleware for an AI-native service economy.
Risks and Limitations
Even so, the proposal is not presented as a cure-all. The source material explicitly notes that Sybil attacks remain possible, and validator honesty still depends on external incentive structures. Onchain hashes can prove that information has not been altered, but they cannot guarantee that the advertised service works as promised or that every reputation signal is meaningful.
That is why ERC-8004 emphasizes layered trust rather than absolute trust. Lightweight reputation may be sufficient for low-stakes tasks, while independent validation is more appropriate for higher-risk interactions. In other words, the design is a framework for making trust more transparent and composable, not a magic solution that eliminates fraud or poor performance.
The Bigger Strategic Picture
At a broader level, ERC-8004 represents another point of convergence between blockchain infrastructure and AI systems. By standardizing identity, reputation, and validation for autonomous agents, Ethereum is positioning itself not only as a financial settlement layer but also as a trust layer for software actors.
With tens of thousands of agents already registered across EVM chains, this is no longer a purely speculative concept. The infrastructure is live, indexed, and expanding. Whether ERC-8004 ultimately becomes the backbone of a major global agent economy or simply serves as an evolutionary step toward later standards will depend on developer adoption, tooling maturity, and governance over time.
For now, however, the direction is clear: Ethereum is giving AI agents a blockchain-native résumé. And by putting identity, reputation, and validation onchain, ERC-8004 is testing whether open markets—not centralized gatekeepers—can decide which autonomous systems deserve trust.