Vitalik Buterin warned AI tools like ChatGPT’s MCP could be exploited, urging human oversight and “info finance” models to safeguard crypto treasuries and decentralized governance.
The rapid rise of AI has brought both innovation and new security risks, and ChatGPT’s latest rollout of the Model Context Protocol (MCP) showed how vulnerable even advanced systems could be.
MCP connects ChatGPT with Gmail, calendars, SharePoint, Notion and other apps to extend utility. Yet, security researchers warned that malicious actors might exploit the system to access private data.
In fact, Security Researcher Eito Miyamura recently revealed how attackers can exploit the system.
She showed how a specially crafted calendar invite containing a jailbreak prompt can trick ChatGPT into reading private emails and sending data to the attacker, without the user accepting the invite.
Although OpenAI runs MCP in “developer mode” with human approvals, experts warn that decision fatigue could cause regular users to inadvertently expose sensitive information.
Responding to the concerns, Ethereum [ETH] Co-Founder Vitalik Buterin emphasized the importance of grounding AI systems in reliable human oversight.
He noted,
“You always have to bootstrap from some ground truth signal that you trust. I think realistically it should be a human jury, where individual jurors are, of course, aided by all the LLMs.”
This highlights the need for combining human judgment with AI capabilities to ensure security and prevent misuse.
He added,
“Also, jailbreaking is not a binary, there’s also lower-grade goodharting, basically the AI-facing equivalent of wearing a suit to look impressive and trustworthy while you defraud people.”
Buterin further advocated for an “info finance” approach, outlined in a previous essay.
In this model, governance frameworks enter an open marketplace where anyone can contribute, while spot checks and final judgments rest with a human jury.
Buterin noted,
“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.”
If looked at carefully, overall, his “info finance” framework mirrors DAO and DeFi governance.
Buterin warned that AI-based fund allocation risks crypto treasuries. He stressed transparency, human oversight, and decentralized accountability.
To provide further clarity, Buterin even emphasized how the Ethereum Foundation guides its fiscal decisions through the newly unveiled Treasury Policy, providing transparency behind each move.
He said the policy moves the foundation from passive ETH holdings to an active, yield-driven approach. It balances financial sustainability with Ethereum’s values of decentralization and privacy.
Additionally, Buterin publicly endorsed Codex, a stablecoin-focused L2 optimized for payments, calling the segment a “large-scale value” and praising Codex’s strategic positioning.
These actions reflect a forward-looking strategy designed to bolster not just Ethereum’s growth, resilience, and adoption, but the broader crypto ecosystem through 2026 and beyond.
