Ethereum Foundation launches zkAPI for anonymous AI payments

The Ethereum Foundation, in collaboration with the Open Anonymity Project, has launched zkAPI on the Ethereum mainnet. The experimental system allows users to pay for AI models and metered APIs without exposing their billing identities. By utilizing zero-knowledge proofs and local client software, zkAPI decouples payment credentials from prompt history, addressing privacy concerns surrounding sensitive AI interactions. However, the system does not provide network anonymity, leaving metadata and prompt content visible to providers.

Launch and origin of zkAPI

• ▪The zkAPI system is live on the Ethereum mainnet and is described as experimental in the zkAPI project's GitHub repository
• ▪The zkAPI system implements a design for zero-knowledge API usage credits published on February 11, 2026, by Ethereum co-founder Vitalik Buterin and Ethereum Foundation dAI team lead Davide Crapis
• ▪The Ethereum Foundation co-developed the zkAPI technology alongside the Open Anonymity Project, an open-source AI privacy effort
• ▪The Ethereum Foundation launched zkAPI on October 1, 2026, a system that allows users to pay for AI models and metered APIs without revealing their billing identities

How zkAPI processes payments

• ▪The zkAPI client software generates a local zero-knowledge proof to show a request is backed by a funded note without revealing which specific deposit covers its cost
• ▪To use zkAPI, users deposit tokens such as ETH or USDC into an Ethereum vault contract, which records the balance as a private note on each user's device
• ▪The zkAPI system publishes a unique serial number called a nullifier with each transaction to detect double-spending attempts without exposing user information
• ▪A zkAPI server verifies a zero-knowledge proof and issues a temporary, short-lived API key with a predefined spending cap for direct communication with an AI provider

zkAPI balance management

• ▪A zkAPI server deducts the cost of actual usage from a user's private balance once the temporary API key it issues expires
• ▪Users can withdraw leftover funds onchain from the zkAPI vault contract even if every zkAPI server goes offline

Integration and use cases

• ▪The primary use cases for zkAPI include AI chat, autonomous agents, blockchain RPC queries, image and video generation, VPN bandwidth, and machine-to-machine payments
• ▪The zkAPI client exposes standard OpenAI and Ollama APIs on localhost, allowing existing applications, editors, and chat clients to integrate without redesigning their code
• ▪The Open Anonymity Project has integrated zkAPI as a payment option within its OA chat application

Limitations of zkAPI privacy

• ▪The zkAPI system does not provide network anonymity, meaning AI providers can still view metadata such as IP addresses and request timing
• ▪AI providers can potentially re-link separate user sessions under zkAPI through prompt content, writing style, or conversation history

Privacy risks of standard AI APIs

• ▪Standard AI API calls require users to associate an API key with an account and a payment method, allowing infrastructure providers to link prompt history to a single user profile
• ▪AI prompts can contain highly sensitive personal details, business strategies, proprietary code, and financial information, making persistent account tracking a privacy risk

Debatable claims

• ▪zkAPI's payment anonymity is sufficient to protect AI user privacy
• ▪Mainstream AI providers should integrate anonymous payment protocols like zkAPI
• ▪Deploying the experimental zkAPI protocol on the Ethereum mainnet is premature
• ▪Anonymous AI payment systems do more harm than good