Money20/20 USA 2026 opens today at The Venetian in Las Vegas. Its Monday panel, “Know Your Agent: Here’s How Banks Are Updating Identity Verification,” brings Skyfire CEO Amir Sarhangi, Firmly CEO Kumar Senthil, PayOS co-founder Johnathan McGowan, VGS CEO Chuck Yu, and J.P. Morgan’s Frank Medrano to the ArrA.I.val Stage for thirty minutes on agent identity in payments. Authenticate 2026, hosted by the FIDO Alliance, opens tomorrow in Carlsbad with a dedicated Non-human Authentication track covering agentic AI, MCP, and API protection. Two major events, same 72-hour window, same structural problem.
The timing is not coincidental. Ten trust-through-defaults incidents in twenty-one days have exposed every layer of the identity stack — framework, library, plugin, appliance, IdP, low-code, utility, browser-side, implementation-layer SaaS plumbing, and MSP credential layer. That pattern, documented across the nine-layer synthesis, the OAuth/SAML bypass wave, and the agent vulnerability trend, has pushed the payments industry past the point of waiting for standards bodies.
Know Your Agent — KYA — is the name the industry has given to the answer. Where OAuth and SAML trust a credential as proof of identity, KYA mandates cryptographic binding of every AI agent request to a registered operator and an authorized user. The question shifts from “do you have a token?” to “are you authorized to act, and for whom?”
Five parallel approaches are competing to solve it. None has emerged as a standard.
Visa launched its Trusted Agent Protocol (TAP) in October 2025, co-developed with Cloudflare. TAP uses RFC 9421 HTTP Message Signatures with Ed25519 cryptographic keys. Every agent request carries signed headers that a merchant verifies against a Visa-operated directory of agent public keys. Two signature tags — one for browsing, one for payment intent — let merchants distinguish window-shoppers from buyers. Verification requires KYB/KYC vetting, published agent identity, and annual recertification. Launch partners include Stripe, Shopify, Adyen, Worldpay, Microsoft, and Coinbase.
Mastercard’s approach splits into two layers. Verifiable Intent, co-developed with Google and open-sourced in March 2026, creates a cryptographically verifiable delegation chain using SD-JWT credentials that bind an issuer, a user, and an agent through key confirmation claims. Eight constraint types — amount bounds, merchant allowlists, budget caps, recurrence terms — are machine-verifiable and tamper-evident. The specification is protocol-agnostic, integrating with AP2, ACP, and other commerce protocols. Worldpay, Fiserv, Checkout.com, and Adyen have signed on. Underneath, Mastercard’s Agentic Tokens, announced in April 2025, issue 16-digit network tokens bound to a real card but scoped to a specific agent, merchant, and consent policy.
Skyfire’s KYAPay protocol takes an open-standards route. JWT-based identity tokens carry three layers: the human principal, the agent platform, and the agent itself. An IETF draft is in progress. The protocol supports both stablecoin settlement and tokenized card payments, with partners including F5, Fastly, Experian, and Akamai.
Google’s Agent Payments Protocol (AP2), announced in September 2025 and donated to the FIDO Alliance in April 2026, uses cryptographically signed mandates — Intent, Cart, Payment — that serve as tamper-evident proof of user authorization. The protocol supports card networks, ACH, real-time payments, and stablecoins on Base. It has 60-plus partners, including Mastercard, PayPal, and Coinbase.
The structural connection to this beat’s coverage is direct. Every one of the ten trust-through-defaults incidents documented since September 26 involved a system trusting an assertion without verifying that the assertion meant what the system assumed. OAuth bearer tokens, SAML assertions, OIDC flows — all relied on the credential being present rather than on the agent being authorized to a specific scope. The KYA protocols attempt to close that gap by making authorization explicit, scoped, and cryptographically verifiable at the point of transaction.
The fragmentation risk is real. Five competing protocols, each with its own directory model, key management, and verification flow, create operational complexity for any merchant or agent platform that needs to work across payment networks. Security teams are navigating a landscape where a single agent may need to support multiple incompatible identity frameworks, potentially creating new seams in the very systems designed to eliminate them.
The legal landscape offers no clarity. A Fenwick legal analysis from April 2026 flags that US regulatory and liability questions around agentic payments remain unresolved. When an agent exceeds its authorized mandate, who bears the loss? Does a cryptographically signed Verifiable Intent credential constitute a legally binding authorization? These questions have no settled answers.
With identity-based attacks doubling in 2026 and 78 percent of organizations reporting identity-related breaches in the past year, the pressure to standardize is mounting. The payments industry is building the trust layer that OAuth and SAML could not. The construction is happening now, in parallel, without a convergence framework. The approaches are sound in isolation — cryptographic binding is structurally superior to bearer-token trust — but the lack of coordination means the industry may need to solve an interoperability problem before it finishes solving the identity problem.
Heath — Forkast T/I/S