How Ethereum Plans to Make AI Private and Scalable with ZK Protocols

Ethereum co-founder Vitalik Buterin sparked fresh discussion around AI and blockchain after sharing his view on how Ethereum can become a true home for artificial intelligence. 

He stressed that the network should not copy existing systems with minor cosmetic changes but instead deliver something fundamentally better through real technological upgrades.

Buterin emphasized the need for payment and reputation services that employ zero-knowledge privacy. He believes that AI services should shield users rather than bind them to identity-based systems.

The adoption of AI is rapidly increasing, with millions of requests going through centralized APIs every day. The requests include personal prompts, financial information, and private business data.

Also Read: Ethereum’s (ETH) Bold 2026 Shift: Key Changes Ahead

Ethereum Foundation Proposes Stake-Based API Access

Davide Crapis, the AI lead at the Ethereum Foundation, spoke about a problem that is becoming increasingly prevalent with the way the internet is currently structured.

Most APIs require you to prove who you are, so every request is tied to an email address, a payment card, or a wallet. Crapis said that AI could make this problem worse.

To solve this, Crapis and Buterin introduced ZK API Usage Credits. This concept relies on stake rather than identity. A user pays the money upfront and can make thousands of API calls without being associated with a physical account.

For instance, a person can pay 100 USDC to query a hosted large language model 500 times. They will be paid requests, but they won’t be able to associate them with the same individual. Another example is paying 10 USDC to query Ethereum’s RPC 10,000 times.

ZK Credits, Refund Tickets, and Slashing Against Abuse

Rate-Limit Nullifiers link anonymity to a financial interest. You can remain anonymous as a legit user, but once you exceed the limit, the secret key is revealed, and you get slashed.

Every request costs a maximum fee, which is paid in advance, and the difference is refunded by the server once the actual cost is determined. These signed refund tickets are privately collected by users to claim back unused credits and more in the future.

The plan also introduces dual staking. There is a stake that enforces the rules of the protocol. Another policy stake can be burned if a user violates the rules of service, such as using restricted prompts.

Also Read: Ethereum (ETH) Tests $2,080 Resistance Amid Rising Bitcoin Dominance Impact