PANews, October 2 — According to official sources, the Ethereum Foundation announced that it has jointly developed zkAPI with the Open Anonymity Project, and it is already running on the Ethereum mainnet. zkAPI uses zero-knowledge proofs to separate payments from identity: users only need to deposit assets such as ETH and USDC into a vault contract on Ethereum once, after which they can authorize AI API usage quotas without revealing their identity. Service providers can see the content of requests but cannot learn the payer's identity, while payment service providers can only see the amount spent and cannot link it to specific requests or users.
zkAPI works as follows: the user's device generates a proof verifying that certain funds are sufficient to cover the current consumption and have not been used before; after the server verifies it, it generates a temporary API key with a quota cap and validity period. The user can then directly use this key to send requests to the AI service provider, which records actual usage and generates a signed credential, and the system ultimately deducts fees from the user's private balance according to actual usage. The Ethereum network only records on-chain operations such as deposits, settlements, and withdrawals, and cannot learn which services specific funds were used to pay for. The solution employs cryptographic techniques such as Groth16, BN254, Poseidon, and Merkle trees, and allows users to withdraw funds directly from the chain if the server stops running.
The Ethereum Foundation stated that zkAPI is initially aimed at AI inference scenarios, because prompts users submit to AI models may involve sensitive information such as health and finances, but the mechanism can also be used for usage-based billing scenarios such as blockchain RPC, image and video generation.