The Architectural Split in Agentic Commerce #
\n\n
The naming convention is deliberate. In HTTP, 401 Unauthorized signals that a request lacks valid authentication credentials. 402 Payment Required is a reserved status code for future use, often associated with digital settlement. By naming its new protocol x401, Proof is signaling an architectural split: x401 handles identity, while x402 handles payment. It is a functional division that acknowledges a reality often glossed over in the rush to automate agentic commerce.
\n\n
On August 4, 2026, the Secure Technology Alliance launched the Agentic Trust and Commerce Forum. The group identified agent identity as one of four core governance questions. Yet, the current landscape of agentic commerce frameworks — including Visa TAP, Stripe ACP, Mastercard Agent Pay, and Google AP2 — is curiously lopsided. These frameworks define what credentials can do, but they remain silent on who is actually behind the agent. They provide the pipes for the transaction, but not the verification of the actor.
\n\n
The Delegation Chain #
\n\n
Proof launched x401 v0.1.0 on June 25, 2026, as an open, issuer-neutral HTTP protocol. Its core mechanism is a 4-node delegation chain: Human (IAL2 verified) to Signed mandate (scoped credential) to Agent (presents via OpenID4VP) to Merchant (verifies independently). Each node requires a verifiable artifact. If any node is removed, the chain breaks. The protocol operates on plain HTTP using PROOF-REQUIRED and PROOF-PRESENTATION headers, completing the handshake through an HTTP 401 challenge, an OpenID4VP presentation, and an HTTP 200 authorized response.
\n\n
This approach relies on OpenID4VP, DCQL, and jwt_vc_json. It is a cryptographic solution to a problem that others are attempting to solve with behavioral data. On August 3, 2026, Visa acquired BioCatch for $2.4B. BioCatch uses behavioral biometrics, analyzing 3,000 data points per session, such as keystroke timing and touch pressure, across 1.8B devices. While BioCatch monitors how a user interacts with a device, x401 focuses on the cryptographic proof of the identity and authorization behind the agent itself. They are fundamentally different mechanisms.
\n\n
Liability and the Regulatory Vacuum #
\n\n
The urgency for such a standard is not merely technical; it is legal. On August 4, 2026, the 9th Circuit Court of Appeals ruled in Amazon v. Perplexity AI that users are liable for their AI agents’ actions under the Computer Fraud and Abuse Act. This decision creates a liability vacuum. Users are now responsible for the actions of their agents, yet there is no industry-standard guidance on how to verify the authorization of those agents. Payment processing systems are technically ready — Paymentology CTO Tim Joslyn has noted that 99% could process agentic payments — but the bottleneck remains the inability to verify agent scope.
\n\n
Pat Kinsel, CEO of Proof, summarized the tension: \”AI is making actions and content effortless to generate. Trust will come from knowing who stands behind them.\”
\n\n
Adoption and the Trust Gap #
\n\n
The protocol is currently in its early stages. It has been submitted to the FIDO Alliance agentic authentication standards workgroup, and the Secure Technology Alliance Forum is scheduled to hold its first in-person meeting on November 17-18, 2026, at the Best Buy corporate campus in Minneapolis. While Proof is a WebTrust-certified CA with FIPS 140-2 Level 3 HSMs and 8,000+ relying parties, x401 remains at v0.1.0.
\n\n
Adoption is beginning, though quietly. Two early adopters have been confirmed: a global payments network is adopting x401 as an identity layer for agent-initiated transactions, and a Tier 1 insurer is evaluating it for delegated account actions. As Gagan Mac, VP Product at Circle, noted: \”x402 answers how an agent pays, x401 answers who it is. Those are the first two questions any agentic transaction has to clear, and now each has an open standard.\”
\n\n
Despite these developments, consumer trust remains a significant hurdle. A Product.ai survey from April 2026 found that only 14% of 1,463 US online shoppers trust AI to execute purchases without verification. Whether a cryptographic protocol can bridge the gap between the legal liability of the user and the skepticism of the consumer remains the central, unanswered question in the development of agentic commerce.