Vitalik Buterin Rejects AI-Run Governance, Advocates Human Oversight Through “Info Finance”

Ethereum co-founder Vitalik Buterin has raised concerns about the growing enthusiasm for AI-driven governance, warning that automated decision-making systems can be easily manipulated and therefore pose a systemic risk to decentralized communities. Writing on X, Buterin argued that delegating governance to artificial intelligence creates a dangerous single point of failure that attackers can exploit with carefully designed prompts.

“If you use an AI to allocate funding for contributions, people WILL put a jailbreak plus ‘gimme all the money’ in as many places as they can,” he remarked. His warning highlights the tension between efficiency promised by automation and the vulnerabilities that arise when human oversight is removed from critical systems.

The comments came in response to a recent demonstration by researcher Eito Miyamura, who showed how ChatGPT could be manipulated to expose sensitive data through its new Model Context Protocol (MCP) tools. These tools allow the AI to connect with external services such as Gmail, SharePoint, and Notion. With little more than an email address, Miyamura was able to coax the system into leaking information, underscoring how easily trust in AI can be weaponized. “Remember that AI might be super smart, but can be tricked and phished in incredibly dumb ways to leak your data,” Miyamura concluded.

Buterin’s proposed alternative is what he describes as an “info finance” model. Instead of giving AI full authority over governance or resource allocation, this approach envisions contributors submitting proposals or models into an open market where humans serve as final reviewers. AI tools can still assist in analysis and filtering, but ultimate judgment rests with people, thereby mitigating the risks of exploit-driven manipulation.

The implications for decentralized autonomous organizations (DAOs) are significant. Buterin noted that many DAOs already struggle with over-delegation, where too much authority is concentrated among a few participants or platforms. Introducing AI without human checks could amplify these risks, leading to governance structures that are both centralized and fragile. By contrast, the hybrid model of info finance maintains efficiency gains while preserving the human element necessary for resilience.

The debate comes at a time when AI systems are increasingly integrated into both public and private decision-making processes. In the crypto sector, where transparency and trust are foundational, the prospect of AI-led governance introduces both opportunities and existential risks. Buterin’s intervention suggests that while AI can enhance processes, it should not displace the critical role of human judgment in systems where accountability matters.

As blockchain governance evolves, Buterin’s stance may shape how projects balance innovation with security. His rejection of AI-run governance in favor of a blended model highlights a recurring theme in the digital asset space: technology can accelerate progress, but without human oversight, it may just as easily undermine it.