TLDR
- The Ethereum Foundation plans to position Ethereum as a trust layer for AI systems.
- The organization will focus on coordination and verification instead of building AI models.
- Davide Crapis said Ethereum can serve as a public verification layer for autonomous agents.
- The foundation supports ERC-8004 to standardize agent identity and trust.
- The strategy promotes privacy, local AI processing, and stronger cryptographic security.
The Ethereum Foundation has outlined a strategy to position Ethereum as a trust layer for AI systems. The organization said it will not compete in building large AI models. Instead, it plans to anchor identity, payments, and verification for autonomous agents.
Ethereum Foundation Outlines Coordination Role for AI Agents
The Ethereum Foundation said it will focus on coordination rather than raw AI computation. Davide Crapis, the AI lead at the EF, presented the plan at NEARCON 2026. He said Ethereum can serve as a “public, governance-less verification layer for AI.”
He explained that AI systems now handle trades, applications, and software tasks. However, centralized control could weaken decentralization and privacy. “If AI doesn’t have the properties we care about, and then we use AI for everything, basically no one has those properties anymore,” Crapis said.
He stated that Ethereum can help agents identify themselves and build trust. The network can also route payments and anchor cryptographic proofs. Heavy computing will remain off-chain on traditional servers.
The EF has supported standards such as ERC-8004 for agent identity and trust. Crapis said developers outside Ethereum have shown interest in these standards. He compared the system to a decentralized review network combined with payment rails.
He said Ethereum can maintain transparent histories for agents. Those records can help assess reputation and past actions. As a result, agents can transact without relying on centralized platforms.
Ethereum Extends Core Principles to AI Security and Privacy
The Ethereum Foundation also aims to bring privacy and censorship resistance into AI systems. Crapis referred to this effort as “Props AI” inside the organization. The program promotes privacy, openness, and security in AI design.
He warned that centralized AI services can build detailed user profiles over time. Queries and usage patterns can reveal personal data. Therefore, the EF supports more local AI processing on user devices.
“We want to create a world where users retain as much data and power as possible,” Crapis said. He added, “We just don’t give it to operators.” The approach seeks to limit unnecessary data transfer to large platforms.
Security forms another part of the initiative. Crapis predicted that AI systems will automate cyberattacks and impersonation. “We will probably see hacks orchestrated by AI,” he said.
He argued that traditional authentication models may fail under AI-driven impersonation. In response, he emphasized the importance of cryptographic keys. Control of a private key provides mathematical proof of ownership.
“In a world where AI is in the wild, we want Ethereum to be the place with the big lock,” Crapis said. He added, “If I have the keys, I still have power.” The EF described the AI program as one of several ongoing priorities.
