Ethereum's ERC-8004: A New Trust Layer for AI Agents
Ethereum Foundation is abandoning on-chain neural network computation in favor of building a decentralized trust environment for autonomous AI agents, powered by the new ERC-8004 identity protocol.
Davide Crapis, head of AI at Ethereum Foundation (EF), has revealed the network's new strategic direction: rather than attempting to run neural network computations on-chain, Ethereum will focus on becoming an independent verification and coordination environment for autonomous AI programs. Crapis shared the vision during a CoinDesk interview at NEARCON 2026.
Why This Matters
The rapid proliferation of artificial intelligence systems threatens to re-centralize internet infrastructure around a handful of tech giants. According to Crapis, this directly undermines the core values of the crypto industry — user sovereignty, privacy, and censorship resistance. Ethereum has no intention of competing with OpenAI or Google in building language models. Instead, the development team aims to prevent corporations from monopolizing the next technological layer.
The EF strategy is split into two key pillars: decentralized coordination of AI agents and user data protection.
Decentralized AI Agent Coordination
Autonomous programs — commonly called agents — need infrastructure for identity, payment exchange, and reputation building. Under the EF's plan, compute-heavy operations will remain off-chain on traditional servers. Ethereum itself will serve as a public registry with no single point of control, enabling agents to discover one another and settle transactions.
Crapis compared the envisioned system to a decentralized version of Google Reviews with built-in payments. To bring this mechanism to life, the team is standardizing an AI agent identity protocol called ERC-8004. According to Crapis, the technology is already gaining traction beyond the Ethereum ecosystem.
Props AI and Local Data Processing
The second strategic pillar operates under the working title Props AI. The problem it addresses is the mass data harvesting conducted by centralized AI services, which build detailed user profiles based on queries.
The alternative approach involves processing data directly on users' devices, bypassing third-party servers entirely. This is designed to eliminate surveillance risks and personal data leaks.
Cryptographic Keys to Replace Passwords
Crapis also raised concerns about escalating cybersecurity threats. Neural networks are becoming capable of convincingly impersonating humans, rendering traditional passwords ineffective. The foundation for a new security paradigm, he argued, should be cryptographic keys that can be mathematically verified.
AI integration has become one of Ethereum Foundation's top priorities. The team acknowledges that algorithms will define the network's next evolutionary phase and is working to position the blockchain as a foundational layer for smart agent operations. Earlier in February, project co-founder Vitalik Buterin published a detailed analysis of potential synergy scenarios between Ethereum and artificial intelligence.
Frequently Asked Questions
What is ERC-8004 on Ethereum?
ERC-8004 is a proposed standard for AI agent identity on the Ethereum network. It is designed to enable autonomous programs to identify themselves, exchange payments, and build reputation through a decentralized registry.
How will Ethereum be used for AI agents?
Rather than running AI models on-chain, Ethereum will serve as a trustless coordination layer for AI agents. The blockchain will act as a public registry for agent discovery, identity verification, and payment settlement without centralized control.
What is Props AI by Ethereum Foundation?
Props AI is a working-title initiative from Ethereum Foundation focused on protecting user data from centralized AI services. The approach processes data locally on user devices instead of sending it to third-party servers.
Why does Ethereum Foundation say passwords are becoming obsolete?
According to EF's AI lead Davide Crapis, neural networks are becoming capable of convincingly impersonating humans, making traditional passwords ineffective. Cryptographic keys with mathematical verification are proposed as the replacement.
Is Ethereum competing with OpenAI or Google in AI?
No. Ethereum Foundation has explicitly stated it will not compete in building language models. Instead, its goal is to prevent corporations from re-centralizing internet infrastructure by providing a decentralized trust layer for AI agent operations.
Read also
AI Audit Uncovers Critical Liveness Bug in Ethereum's Nethermind Client
Octane Security's AI discovered a high-severity vulnerability in the Nethermind execution client that could have halted block production for 38% of Ethereum mainnet validators. The Ethereum Foundation awarded a maximum $50,000 bounty.
How ERC-8004 and x402 Standards Are Turning AI Agents Into Market Participants
Blockchain standards ERC-8004 and x402 are building the infrastructure for autonomous machine payments, challenging Big Tech's closed ecosystems and the attention economy model.
Bitcoin Down 2.5% Weekly: Jane Street Accusations & 7 Ethereum Forks
Bitcoin lost ~2.5% over the week amid macro shocks and geopolitical tensions. Jane Street faced market manipulation allegations while Ethereum unveiled an ambitious seven hard fork roadmap through 2029.
TON Wallet Introduces Yield Vaults for BTC, ETH, and USDT Directly in Telegram
TON Wallet has launched yield vaults for BTC, ETH, and USDT directly within Telegram, offering up to 18% APY on stablecoins through partnerships with Morpho, TAC, and Re7.
Bitcoin Hits $70,000 as Iran Ceasefire Talks Boost Risk Appetite
Bitcoin surged 4% to test the $70,000 level on April 6 amid reports of ceasefire negotiations between the US, Israel, and Iran. The derivatives market, however, sends mixed signals.
Bitcoin Rebounds to $70,000 as Leverage Drops and ETF Inflows Continue
BTC recovered above $70,000 on March 10, erasing weekend losses. Spot ETFs attracted $568M in weekly inflows while the estimated leverage ratio on Binance fell sharply from 0.198 to 0.152.
