Vitalik wrote a proposal teaching you how to secretly use AI large models

Original Title: “Vitalik wrote a proposal teaching you how to secretly use AI large models”

Original Compilation: Deep Tide TechFlow

The whole world is talking about AI, and the voices about cryptocurrency have quieted down quite a bit on the timeline.

At the same time, ETH has been hovering around 2000 for almost two months, and it seems that not many people care about what Vitalik says or does.

However, I recently browsed his X and found that the influence of AI is not limited to us. A significant portion of what he has posted in the past month is related to AI, and it goes into the level of technical solutions.

The most noteworthy is a proposal he co-published with Davide Crapis, the AI head of the Ethereum Foundation, on ethresear.ch on February 11, titled "ZK API Usage Credits."

In one sentence: Use zero-knowledge proofs to allow you to anonymously call AI large models.

Currently, whether you use ChatGPT or call Claude's API, there is only one payment method:

Register an account, bind an email, bind a credit card.

Every conversation, every prompt, the platform knows it was you who sent it. What you asked, when you asked it, how many times you asked, all tied to your real identity.

Vitalik and Crapis's proposal offers another path.

1. Users deposit a sum of money into a smart contract, for example, 100 USDC.

2. The contract will register this deposit in a cryptographic list on the chain. After that, every time you call the API, you do not need to present your identity, just generate a zero-knowledge proof.

3. It can prove two things to the service provider: you are on the list, and your balance is sufficient. But the proof itself does not reveal which one you are on the list.

The service provider can receive payment and prevent abuse, but will never know who you are.

You can understand this proposal as Vitalik's belief that in the AI era, users should not have to give up their identity to use an AI tool.

This proposal is still in the research stage and has a distance to go before implementation; large model providers may not agree to such a method. Meanwhile, the comments section of the proposal is filled with rebuttals and doubts, arguing that AI model providers will always find a way to know your real identity.

However, I believe the significance of this proposal does not solely lie in whether it can be implemented.

Privacy has been something Vitalik has worked on for ten years. From early support for Tornado Cash to promoting zero-knowledge proofs as a core technical route for Ethereum, this line has never been broken. It's just that in the context of the cryptocurrency industry over the past few years, privacy has always lacked a sufficiently large story to support it.

AI has filled this story. When you talk to large models more than you talk to anyone else every day, privacy is a real need.

Vitalik Embraces AI

From February until now, a considerable portion of what Vitalik has posted on X is related to AI, with a density that does not seem casual.

Yesterday he posted a long thread saying that he recently attended a cryptography conference where people cared about privacy, open source, and anti-censorship... but had no feelings for blockchain.

Among that group of people, he conducted a thought experiment:

Forget about "we are the Ethereum community," and think from scratch about where Ethereum is most useful.

His conclusion is that the fundamental value of Ethereum is as a bulletin board. A place where anyone can write, anyone can read, no one can modify, and no one can delete.

In the context of AI, this might be the most important thing Vitalik has said in the past two years.

We are entering an era of generating infinite cheap content. Text, images, videos, identities, AI can produce them in bulk. When everything can be forged, what becomes scarce?

These questions ultimately point to the same place: a public, persistent, irreversible data layer. And a record that no one can tamper with is precisely what Ethereum can do.

In the past two years, the doubts faced by Ethereum can be summarized in one question: what do you have that others cannot replace?

Looking back now, Vitalik did not directly answer this question.

However, the Ethereum Foundation has done a few inconspicuous things over the past year: formed a privacy team of 50 people, established a privacy research cluster of nearly 50 people, released the Kohaku privacy framework, and specifically appointed an AI head; in the 2026 roadmap, institutional-level privacy and faster transaction confirmations are listed as top priorities.

Looking back at his intensive output over the past month, it is basically discussing Ethereum's privacy and efficiency issues in the context of AI.

I think Vitalik is betting on one thing: the more powerful AI becomes, the more rigid the demand for privacy and verification infrastructure will be. Whether Ethereum can meet this demand is another matter, but he has clearly chosen his table.

ETH is still hovering around 2000. Most people still do not pay much attention to what he has been saying recently.

But perhaps in a few years, looking back, what should have been paid attention to is this period of time.