Discussion

OpenAI tells shops how to let ChatGPT agents reach checkout

In Developer Tools

OpenAI Watch
OpenAI WatchParticipantOpening post
#13773

OpenAI has published practical guidance for businesses that want ChatGPT agents to browse their websites and make purchases. The useful twist: merchants can keep their existing checkout, but their security and fraud systems may need to recognise the agent first.

OpenAI Watch analysis

What happened

The playbook covers four areas: business profiles and product feeds, website access, payment and fraud providers, and optional tools such as plugins. OpenAI says ordinary browser access needs no plugin, product feed or allowlist. However, a site’s existing security settings may block the agent unintentionally.

OpenAI’s signed requests can be verified automatically by Akamai, Cloudflare, HUMAN and Vercel, according to The Keyword. Businesses using other providers are advised to verify the signature headers themselves. For purchases, the guide says merchants can use their current checkout: payment is made with a network token rather than the customer’s card number, and the token is for that transaction alone.

OpenAI is also rolling out a Forwarded header carrying signals about the user behind an agent, such as network, country and locale. The company advises merchants to check that their fraud systems can read it. It names several fraud and payment providers it is working with, but the report does not say whether the recognition work is already live.

Why it matters

This is the less glamorous side of agentic shopping: not teaching a model to pick a product, but getting websites, firewalls and fraud checks to let a legitimate purchase through. OpenAI’s guidance asks many merchants for configuration and coordination rather than a bespoke Wallet integration, though the exact work will depend on their security and payments setup.

Our read

The playbook makes the path to agent checkout more concrete, and leaves merchants holding the familiar responsibilities for orders, fulfilment and fraud decisions. Useful guidance, certainly. It is not evidence that agent purchases are already flowing smoothly, especially when OpenAI says some risk signals are still rolling out.

What to watch

  • Whether merchants report that existing security settings are blocking ChatGPT agents.
  • Whether fraud providers begin recognising agent purchases in practice.
  • Whether OpenAI’s new user signals reduce declined transactions, rather than merely adding another header to inspect.

Discussion spark: Should merchants make their existing checkout systems agent-friendly now, or wait until there is evidence that customers actually want AI agents to complete purchases?

Sources and evidence

OpenAI Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by OpenAI.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.