← The Polylog Crypto Intelligence Brief.jpg)
Morning Edition · Saturday, October 3, 2026Published at 1:57 AM EDT · New York
Ethereum Foundation Ships Anonymous Payments for AI Model Access on Mainnet
zkAPI lets a user prove they have sufficient funds and receive a temporary, spending-capped key without revealing their identity. Miden launched a private stablecoin wallet the same week.
The Ethereum Foundation, working with the Open Anonymity Project, launched zkAPI on Ethereum mainnet this week, a payment rail that charges users for application programming interface (API) calls without learning who they are. A user deposits Ether or USD Coin (USDC) into a vault contract. The user's own device then generates a zero-knowledge proof that the deposited funds are sufficient for the request and have not already been spent, and the server returns a temporary API key with a spending limit and an expiry. The Block reports that the service provider sees the request but not the payer, while the payment processor sees only amounts.
The system uses Groth16 proofs and Merkle trees, and implements a design that Ethereum co-founder Vitalik Buterin and Davide Crapis, who leads the Foundation's decentralized artificial-intelligence work, published in February. Paid inference from artificial-intelligence models is the first use case. The same construction applies to remote procedure call access to blockchain nodes, media generation, virtual private networks and other machine-to-machine services where the buyer currently has to supply an identity to get a key.
Two limitations come from the design itself. Groth16 requires a trusted setup, and the privacy a user actually obtains depends on how many other depositors share the vault, because a small anonymity set narrows the field of possible payers.
Miden pursued a related goal the same week, announcing a stack it calls PriFi for private stablecoin finance, built around USDCx, a token it pegs one-to-one to USDC. Its self-custodial wallet, Bread, executes and proves transactions on the user's device and publishes only a 32-byte commitment to the chain, with testnet builds on iOS, Android and Chrome.
Veracity: Corroborated
88/100- If true, who benefits
The Ethereum Foundation's privacy program and the zero-knowledge engineering sector gain a paying use case outside trading, while artificial-intelligence model sellers gain a way to collect revenue without storing customer identity and payment processors lose data they currently monetize.
Part of a tracked trend
Privacy Chains Pivot From Niche to Institutional Pitch
Over 3-6 months, confidential execution reframes as a prerequisite for serious/institutional on-chain use, with privacy L2s shipping live networks and contesting who controls confidentiality.
Townsquare
Start a discussion in Townsquare.
More from this edition
- Blast Is Closing Its Ethereum Layer-2 and Has Given Users Until October 26 to Withdraw
- Attacker Forged a Safe Authorization Check to Drain 114 Ether From an Aave Looping Module
- SEC Proposes a 760-Page Custody Framework That Would Let Advisers Hold Crypto Themselves
- Aztec Still Awaits a Fix for a Proving Flaw That Could Accept Invalid Transactions
- Tether and Circle Have Added About $200 Billion of Treasury Exposure in Five Years
- Community Bankers Sue the Comptroller of the Currency Over Crypto Trust Charters
- BNB Chain Passes $1 Billion in Tokenized Stocks and Funds, Taking Share From Ethereum
Comments
1Oct 3, 5:57 AM · edited
Because the temporary key is unlinkable to the payer, providers lose identity based rate limiting, so the spending cap and per call price become the only Sybil defense against abuse.