Ethereum zkAPI Privacy Tool Separates Payments From AI Usage

The Ethereum Foundation and Open Anonymity Project have introduced zkAPI, a system designed to enhance privacy for AI and other metered services. Launched on October 1 on the Ethereum mainnet, zkAPI allows users to prepay for services without linking their billing identity to each request. The system leverages zero-knowledge proofs and disposable API keys to separate the payer from the service provider.

Users deposit ETH, USDC, or other supported cryptocurrencies into an Ethereum vault. These funds are then represented by private notes that can be spent without revealing the source of the payment. The system generates zero-knowledge proofs to verify that a funded note covers the requested spending without exposing which note is used. This ensures that users remain unlinkable as long as they stay within their balance.

After payment authorization, zkAPI creates a temporary, short-lived API key with a spending cap. This key is used to send prompts directly to the AI provider, ensuring that the payment server never sees the content, and the AI provider never learns the billing identity behind the key. The system also allows users to reclaim their funds if the zkAPI servers become unavailable.

However, zkAPI does not provide complete anonymity. The AI provider still receives the prompts, and network information such as IP addresses can potentially link different sessions. The Ethereum Foundation acknowledges these limitations and advises users seeking stronger network anonymity to use tools like Tor. The system is currently experimental but has potential applications beyond AI, including blockchain RPC queries and machine-to-machine services.