Ethereum Co-Director Unveils AI Governance Revolution Plan
Tomasz Stańczak reveals a five-step blueprint for LLM-driven Ethereum governance, positioning the network to beat rivals in the AI-powered blockchain race.
The Shift Towards AI Governance
Ethereum’s governance could soon run on artificial intelligence. Tomasz Stańczak, co-director at the Ethereum Foundation, posted on X a detailed five-step plan to make Ethereum the first blockchain driven by large language models (LLMs).
The Five-Step Plan
The first step involves validator operators handing decision power to AI agents. These agents would handle network upgrade approvals and parameter settings. Stańczak argues this shift mirrors Ethereum’s early advantage as the first proof-of-work chain – being first matters in the AI race too.
The proposal comes at a time when blockchain governance faces mounting complexity. According to Stańczak on X, Ethereum holds a natural edge because LLMs were already trained on existing Ethereum specs and transparent governance records.
Step Two: Empowering EIP Authors
Step two pushes EIP authors to use LLMs for creating and submitting Ethereum Improvement Proposals (EIPs). The third step extends AI review tools to EIP editors. All Core Developers would then rely on LLMs to moderate meetings and vote on EIP inclusion under step four.
Ensuring Effective Participation
Priority one is ensuring agentic participation in EIP submission works smoothly. EIP editors need proper tooling for AI review of all proposals, Stańczak stressed in his X post.
He wants real-time ACD moderation with AI support. The system would connect to chat, analyze discussion content live, and offer suggestions as conversations unfold.
Building a Cross-Client Core Dev Team
The proposal also calls for a cross-client core dev team. This group would work exclusively on an AI-generated client codebase built only from specs – no human coding involved.
Emphasizing Formal Verification
Such a client must be fully formally verified and test-covered, according to Stańczak. Development would run parallel to existing codebases until the AI-generated version becomes canonical.
The timing isn’t random. Ethereum already has thousands of hours of recorded All Core Dev calls. EIP processes are documented. Open discussions are archived. All of this serves as training data for LLMs.
A Vision for AI Governance
Stańczak’s plan treats AI governance as an infrastructure upgrade, not an experiment. The five steps build on each other: validators delegate to agents, authors use LLMs, editors review with AI, developers vote through LLMs, and client teams generate code automatically.
Between tooling coordinators and the dAI team, implementation appears already underway. The proposal acknowledges this reality rather than starting from scratch.
Competitive Edge and Challenges
Being first to achieve LLM-driven governance gives Ethereum the same edge as proof-of-work once provided. Other chains will follow, but Ethereum’s existing specs and governance transparency create a moat that’s hard to replicate.
The five-step framework doesn’t promise easy execution, though. Real-time AI moderation of technical discussions requires sophisticated natural language processing. Generating formally verified client code from specs alone pushes current AI capabilities to the limit.
Conclusion: A Transformative Path
Still, Stańczak’s post suggests confidence in the path. The pieces are in place – teams hired, infrastructure expanding, governance records ready for training. What remains is execution across a five-step process that could redefine how blockchains evolve.
Ethereum’s transparent governance history becomes its biggest asset in this transition. Every past decision, every ACD call, every EIP debate – all available for LLM training. Competitors without this documentation start from behind.
The proposal positions AI not as a replacement for human judgment but as a tool for better governance at scale. Validators still decide whether to accept agent recommendations. Authors still craft EIP concepts. Editors still approve submissions. AI amplifies their capacity.
If successful, Ethereum’s governance could process upgrades faster while maintaining decentralization. The canonical AI-generated client would serve as a reference implementation, reducing inconsistencies across client teams.
