When an AI agent pays for something on your behalf, should it prove who it is using brand-new rules, or the login standards the web already trusts?
Should payments infrastructure adopt agent-specific protocols or reuse existing identity standards?
This explores whether letting AI agents pay for things calls for new protocols built just for agents, or whether the identity and authorization standards the web already runs on (OAuth, OpenID Connect, HTTP signatures) can be stretched to cover them.
This explores whether agent payments need new protocols built just for agents, or whether the web's existing identity standards can be stretched to cover them. The corpus doesn't treat this as either/or. The stronger answer is that the agent-specific protocols mostly sit on top of existing standards. The real design question is which parts of the problem are actually new.
The case for reuse is strong. NIST's NCCoE argues that OAuth 2.0/2.1, OpenID Connect, and SPIFFE/SPIRE already cover most of what agents need: identification, authorization, delegated access, logging, and tracking where data goes. It also points out that MCP is already built on these standards Can existing identity standards work for AI agents?. The 'agent-specific' protocols look less new up close too. Visa's Trusted Agent Protocol, built with Cloudflare, is an extension of the existing HTTP Message Signatures standard. The new part is what goes inside the signature: intent, consumer recognition, and payment data Can cryptographic signatures help merchants identify trusted AI agents?. The same pattern shows up outside payments. Agent coordination standards that wrap and bridge existing protocols get adopted, while ones that try to replace them stall Should coordination protocols wrap existing systems or replace them?.
There is a real gap, though, and it's one you might not expect. Existing standards are good at letting a user safely hand credentials to an agent. They don't give the merchant a way to recognize that an agent is the one shopping. Allglenn's reading of Amazon blocking Meta's Muse agent makes this concrete. Vaulted credentials keep the consumer's secrets safe, but that is a one-sided control. Without a standard, legible way for agents to identify themselves, retailers fall back on terms-of-service blocks instead of negotiated access Can credential vaulting alone solve merchant-facing agent identification? What identity standards do shopping agents actually need?. McKinsey identifies a second gap in fraud detection. Fraud engines are trained on human signals like typing patterns, session behavior, and explicit intent, and those signals disappear when software makes the purchase. Its proposed fix is cryptographically signed mandates that record what the human actually authorized Does agent-initiated commerce break existing payment infrastructure?.
That's why the field is building a stack of protocols, not one winner. Boboev describes ACP, AP2, UCP, TAP, and Web Bot Auth as a ladder of layers: discovery, application logic, governance, and identity. Merchants combine them, and each layer absorbs a different level of fraud risk Do agentic commerce protocols compete or complement each other?. The IMF draws the most important line in that stack. The agent's probabilistic decision-making ('should I buy this?') has to stay separate from deterministic authorization and settlement, because settlement needs predictability and legal enforceability that a reasoning model can't guarantee Should AI agents make payment decisions independently?. By that logic, the parts that move money should stay boring and reuse existing rails. The new protocol work belongs at the boundary where agent intent gets turned into something verifiable. Simon Taylor's sequence for the rails themselves fits this: virtual cards first because their controls and acceptance already exist, then stablecoin-settled cards, and native wallets only later Will agents adopt cards before stablecoins for payments?.
Two more points complicate the picture. First, trust may come from brands as much as from protocols. In a Visa-commissioned survey, only 23% of US adults trusted AI to make payments, but 61% trusted Visa to handle agent-initiated transactions Does naming a payments brand increase trust in AI transactions?. That gives incumbent networks a reason to brand their own agent protocols instead of deferring to neutral standards bodies. Second, the UK's competition regulator (the CMA) warns that if agents can't move a user's data between ecosystems, they end up locked into incumbent platforms Should data portability be infrastructure for agent-driven markets?. So 'which protocol' is partly a question about who controls the gate. The corpus doesn't yet show which approach wins in practice, and most of these sources are proposals and commentary, not deployment data.
Sources 11 notes
NIST's NCCoE concept paper argues that OAuth 2.0/2.1, OIDC, and SPIFFE/SPIRE should be applied to AI agents rather than developing new protocols. The paper identifies five priority areas—identification, authorization, access delegation, logging, and data flow tracking—and notes that MCP already embeds these standards as its foundation.
Visa developed Trusted Agent Protocol with Cloudflare to enable merchants to authenticate AI agents through agent-specific cryptographic signatures carrying intent, consumer recognition, and payment data. The protocol builds on HTTP Message Signatures and aims to bridge the gap between AI-assistant adoption and merchant trust.
Research shows that agent coordination standards achieve adoption by composing existing protocols like MCP and DIDComm under a shared substrate, rather than competing to replace them. Bridging lets value accrue incrementally without forcing ecosystem-wide rewrites.
Secure credential storage is a unilateral consumer-security control that prevents agent exposure of secrets, but it cannot substitute for a visible, retailer-legible self-identification protocol that both the account holder and merchant must agree to separately.
Amazon's block of Meta's Muse reveals that AI agents need standardized ways to identify themselves and prove authorization to retailers—separate from consumer-side credential security. Without such protocols, retailers enforce access through terms-of-service blocks rather than negotiated standards.
Show all 11 sources
McKinsey argues that payment gateways and fraud engines built around explicit human identity and intent face significant strain when autonomous agents initiate transactions. The firm proposes cryptographically signed mandates as a replacement for human-observable signals fraud systems traditionally relied upon.
Boboev argues ACP, AP2, UCP, TAP, and Web Bot Auth each solve different layers—application, governance, discovery, and identity—creating a capability ladder merchants stack rather than choose between, with each layer absorbing different levels of fraud risk.
An IMF analysis argues that agent-mediated payments must split probabilistic AI decision-making (Layer 1) from deterministic authorization and settlement (Layers 2–3) because core payment infrastructure requires predictability and legal enforceability that probabilistic reasoning cannot guarantee.
Fintech commentator Simon Taylor argues agent payments follow a three-stage sequence driven by increasing complexity: virtual cards leverage mature controls and universal acceptance, stablecoin-settled cards solve cross-border settlement delays, and native wallets enable multi-agent permission-less structures. The progression reflects different tools for different jobs rather than competition.
A Harris Poll survey found 72% of US adults use AI assistants, but only 23% trust AI to execute payments. When Visa's name was attached, 61% reported trust in the company to handle agent-initiated transactions—suggesting consumers prefer a recognized intermediary over AI alone.
The CMA frames data portability as a precondition for competitive agentic markets, warning that without it, agents cannot act effectively on consumer behalf and switching becomes blocked by lock-in to incumbent ecosystems.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Infrastructure for AI Agents
- Payments in the Age of AI Agents
- The agentic commerce opportunity: How AI agents are ushering in a new era for consumers and merchants
- How Agentic AI Will Reshape Payments
- From Product Search to Preference Articulation: The Economics of Agentic Commerce
- Authenticated Delegation and Authorized AI Agents
- The Coasean Singularity? Demand, Supply, and Market Design with AI Agents
- Deep Dive: The hidden liability of agentic commerce