Proof launched x401, an open protocol for AI agent authorization

Proof launched x401, an open protocol that lets a website make a request to an AI agent. The agent must give the identity of the person it represents and the things that person approved. Proof says it works with assistants that use the Model Context Protocol (MCP), such as Muse from Meta, Grok and Claude. Developers can use it at this time.

How it works

The agent identity wallet holds the verified legal identity of a person. The person gives approval on their device. The approval sets the things that the agent can do.

The agent carries the request. The wallet signs it.

The business that gets the request can check the identity. It can also check the cryptographic signature on the transaction. If the agent changes a signed term, the signature is not correct.

What is in the launch

• The wallet, which shows the data and the terms of the request
• x401, an open HTTP protocol for this check
• A production MCP server for AI assistants

Why now

Weeks before this, Amazon blocked Muse from its marketplace. Proof says that no reliable procedure exists to find who an agent represents.

In September, FinCEN and federal regulators gave guidance on digital certificates. In some conditions, a bank can use them to make sure of the identity of a customer.

How to start

A website can request verified identity claims with no sales procedure. It can pay for each claim with x402.