FRVKTUREFRVKTURE
Notes

Who pays the chargeback when an AI agent buys the wrong thing

20 November 2026 · Agentic commerce · 6 min read

Something new is going to start showing up in chargeback reports, and most merchants will not see it coming.

AI agents are buying things now. Not browsing. Buying. A customer tells an assistant to reorder, to compare, to subscribe, and the assistant completes the checkout. The plumbing for this is published and in production: an open protocol for agentic checkout and a delegated payment specification that lets an agent present a payment credential on a customer's behalf [1][2], with the large commerce platforms wiring into it [3].

Here is the part that matters if you take cards.

The merchant stays the merchant of record

In every framework published so far, the business selling the goods remains the merchant of record. The agent platform does not become the merchant, does not become the seller, and does not take on the settlement relationship.

So the refund is still yours. And the chargeback is still yours.

Meanwhile the card networks have moved on identification, not on liability. Visa's agentic work defines how a trusted agent is recognised and how an agent initiated transaction is signalled through the authorization message [4]. That is a real and useful step, because without it an agent order looks like a bot to your fraud stack. But recognising a transaction type is not the same as writing a new dispute right for it, and no network has published one that shifts the loss away from the merchant.

Which means that when a customer says I never told it to buy that, the case gets argued under reason codes written for a human sitting at a checkout.

Your evidence file just got thinner

Now think about what you normally win a dispute with.

Half of that gets thin when the buyer was never looking at a screen. The IP belongs to an agent platform's infrastructure. There is no device, no session, no dwell time. The behavioural signals your processor's fraud tools were built to read are simply not generated.

For a business already being watched on its dispute ratio, that is not a technology story. It is an underwriting story. Ratios move first. Reserves follow. If you want the arithmetic on why your bank reacts before you hit any published threshold, we wrote that up in what is my chargeback limit, and who actually sets it.

What has to replace it

The merchants who come through this clean will be the ones logging something different. Three fields, captured at the time of the order, not reconstructed six weeks later when the case arrives:

That record is what replaces the device fingerprint. It is not a speculative requirement. It is the only material you will have to work with.

One question worth asking before the next busy season

Ask your processor, in writing: can an agent initiated order be identified in my transaction data, and is the agent signal retained where I can pull it for a dispute response? A surprising number of stacks receive that signal and drop it. Finding out in January is cheap. Finding out during a dispute is not.

What not to do

Blocking agent traffic is the obvious reflex and the wrong one for most merchants, because it blocks real customers with real intent and real money. Agent originated orders tend to be high intent: somebody explicitly told the assistant to buy.

The other wrong move is assuming your fraud tooling will sort it out. Rules tuned on human behavioural signals will either decline good agent orders or wave through bad ones, because the signals they score are absent. That is a tuning project, not a setting.

Nobody has to solve this this week. But if you sell subscriptions, digital goods, or anything that bills on repeat, it is worth asking your processor one question before the next busy season.

What does my evidence file look like when the buyer is not a person?

Questions people actually ask

Who is liable when an AI agent makes a purchase the customer disputes?

In every agentic commerce framework published so far, the merchant remains the merchant of record. The refund obligation and the chargeback liability stay with the business that sold the goods, not with the agent platform that completed the checkout. The networks have added ways to identify an agent initiated transaction, but they have not published a separate dispute right that moves liability to the agent.

Can a customer dispute a purchase their AI assistant made?

Yes. A cardholder can dispute any transaction on their account, and an agent initiated purchase is argued under the existing reason codes, which were written for a human at a checkout. A claim of "I never authorized that" is handled as an unauthorized or not as described case, with the merchant supplying the compelling evidence.

Does an agent transaction count toward my chargeback ratio?

Treat it as though it does. The monitoring programs count disputes and fraud reports against settled transactions without carving out agent initiated ones, so an agent driven rise in disputes shows up in the same ratio your acquirer is already watching.

What evidence should a merchant log for agentic transactions?

Which agent transacted and on whose behalf, the scope of what the customer authorized the agent to do, the specific cart the agent approved, and the timestamps for each. That record is what has to replace the device fingerprint and browsing session you would normally submit.

Should I block AI agents from checking out?

For most merchants that is the wrong trade, because it blocks real demand. The better position is to accept agent traffic and instrument it, so an agent initiated order is identifiable in your own records before a dispute arrives rather than after.

Sources and further reading
  1. OpenAI, commerce and agentic checkout documentation. The production guidance for agent completed checkout, including where the merchant of record sits.
  2. Agentic Commerce Protocol. The open protocol and the delegated payment specification that lets an agent present a credential on a customer's behalf.
  3. Shopify. Platform level implementation of agentic checkout, with the store remaining the seller.
  4. Visa Intelligent Commerce and the Trusted Agent Protocol. How an agent is recognised and how an agent initiated transaction is signalled. Identification, not a transfer of liability.
  5. Google, commerce developer resources. The parallel protocol work, with the same merchant of record position.
  6. Visa, merchant resources. Dispute reason codes and compelling evidence requirements, which have not been rewritten for agent purchases.

If you bill on repeat and want this looked at before it shows up in a report, ask us.

Talk to us

FRVKTURE Payments is an independent placement agent for payment processing and related services. We are not a payment processor, a bank, or a card network. All services are provided through established, registered processing partners and their sponsor banks. Agentic commerce standards and card network rules are moving quickly, and everything here reflects what was published at the time of writing. Nothing here is legal or compliance advice.