Vitalik Buterin: AI in Governance Is a Trap 🚨

[etfs]

Ethereum co-founder Vitalik Buterin has warned against letting AI manage governance or treasury funds in crypto projects.

His reasoning:
“If you use an AI to allocate funding, people WILL find jailbreak exploits like ‘gimme all the money’ and drain the system,” he wrote on X.

Why it matters:

DAOs and DeFi teams are testing AI for governance and resource allocation

But AI models can be easily tricked, jailbroken, or phished

One bad exploit could wipe out treasuries or skew governance votes

Buterin says the community must be careful not to outsource trust to fragile AI systems.

[Nahid Hossen]

Vitalik is right—AI is powerful, but one clever jailbreak could drain a DAO overnight. Governance needs checks, not blind automation.

[Abdul Khan]

The idea of “AI treasurers” sounds futuristic, but in practice it’s a huge attack surface. Crypto can’t afford FTX-level blowups from AI exploits.

[JanEmil]

You can always blame AI if something goes wrong

[Rimon Khan]

Prediction markets and human spot-checks still feel safer than handing DAO funds to a model that can be tricked by a single prompt.