Reports Suggest OpenAI Is Pulling Back on Native Checkout. Here's What It Means for AI Distribution.

AI distribution is the practice of making products discoverable, quotable, and purchasable directly inside AI assistants like ChatGPT, Claude, and Gemini. In March 2026, multiple reports emerged suggesting that OpenAI is scaling back native checkout inside ChatGPT and shifting toward merchant-owned apps. If confirmed, this would be the most significant structural decision yet about how AI commerce will work.

In early March 2026, The Information reported that OpenAI was pulling back on its native checkout ambitions inside ChatGPT (The Information, March 2026). The feature, branded “Instant Checkout,” had launched just six months earlier with Stripe integration, PayPal partnership, and access to over one million Shopify merchants. The vision was simple: let users discover and buy products without leaving the conversation.

The early signals suggest it hasn’t worked as planned. And the reasons why reveal something important about how AI distribution will actually function.

What we know so far

OpenAI launched Instant Checkout in September 2025 as part of the Agentic Commerce Protocol (ACP), co-developed with Stripe. The premise was that ChatGPT could become a full commerce platform: discovery, comparison, checkout, all in one conversation. Shopify merchants were charged a 4% transaction fee on top of their existing Shopify fees (PYMNTS, January 2026). Etsy sellers went live in February. Instacart, DoorDash, and Target launched their own ChatGPT apps around the same time.

By March, the early results appeared discouraging. According to The Information, only about 12 of Shopify’s million-plus eligible merchants had actually integrated with native checkout. Users were researching products in ChatGPT but completing purchases elsewhere. Forrester’s ConsumerVoices survey, published the same week, found that completing a purchase inside an AI platform is the least-adopted use case among regular users (Forrester, March 2026).

The technical challenges were reportedly severe. As of February 2026, OpenAI had not built a system for collecting and remitting state sales tax in the U.S. Real-time inventory synchronization across millions of SKUs proved unworkable. Fraud prevention, refund handling, and order management were all reportedly unresolved.

Reports state that an OpenAI spokesperson said: “We are evolving our commerce strategy within ChatGPT to better meet merchants and users where they are. Instant Checkout is transitioning to apps, where purchases can occur more seamlessly” (Lengow, March 2026). OpenAI has not explicitly confirmed that native checkout is being discontinued, but the direction of travel appears clear.

Why native checkout was always going to be difficult

Whether or not OpenAI formally kills Instant Checkout, the structural problems with platform-owned checkout in AI are real. Understanding them matters for anyone thinking about AI as a distribution channel.

1. Users don’t trust AI with transactions (yet).

Forrester’s data is unambiguous: consumers use AI platforms for research and comparison, not purchasing. This tracks with established behavioral patterns. Shoppers have saved payment methods, order histories, and loyalty programs with retailers they already trust. A new checkout flow inside a chat window doesn’t compete with one-click ordering on a site where you’ve bought 50 things before.

This does not mean AI commerce is dead. It means the transaction layer needs to live where trust already exists: inside the merchant’s own experience.

2. Platform-owned checkout creates misaligned incentives.

OpenAI’s 4% transaction fee put it in direct competition with Google and Microsoft, neither of which currently charges Shopify merchants additional fees for purchases through their AI interfaces (Retail Brew, January 2026). For merchants already paying Shopify’s processing fees, adding a 4% toll to a channel with unproven conversion rates was a hard sell.

Marketplace Pulse described it plainly: the 4% fee confirmed marketplace economics (Marketplace Pulse, 2026). OpenAI was trying to become a marketplace intermediary, not a distribution platform. Merchants noticed.

3. Complex products were never going to work as “buy now” buttons.

Instant Checkout was designed for simple retail: a pair of shoes, a kitchen gadget, a candle. Products where the entire purchase decision can happen in a few messages. But the highest-value AI distribution opportunities are in industries where purchase decisions require personalization, context, and compliance: insurance, financial services, B2B SaaS, travel, healthcare.

You cannot sell an insurance policy with a checkout button. You need a conversation that understands the customer’s situation, generates a personalized estimate, explains coverage, handles regulatory disclosures, and captures the lead for follow-up. That’s not a checkout flow. That’s a storefront. This is precisely why we built WaniWani as AI distribution infrastructure rather than a checkout layer: the products that benefit most from AI distribution are the ones that require a real conversation, not a cart.

This has happened before

If OpenAI does step back from native checkout, it would not be the first platform to try owning the transaction layer and retreat.

Google launched “Buy on Google” in 2015, letting users purchase products directly from Google Shopping results without visiting a merchant’s website. Google recruited sellers aggressively, attended conferences, and by 2020 had even reduced commission fees to zero. Despite all of this, only about 8,000 sellers integrated (Marketplace Pulse). Google killed the feature, and all shopping results now link to merchants’ external websites.

The pattern is consistent: discovery platforms are good at discovery. The transaction belongs with the merchant.

Where OpenAI appears to be heading

Regardless of what happens with Instant Checkout specifically, the broader direction is becoming clear.

The company is doubling down on discovery. ChatGPT Shopping Research, launched earlier in 2026, is powered by a version of GPT-5 mini trained specifically for shopping tasks using reinforcement learning. OpenAI’s internal evaluations show it reaches 52% product accuracy on multi-constraint queries, compared to 37% for standard ChatGPT Search (OpenAI, 2026).

For transactions, the emerging model is merchant-owned apps inside ChatGPT. Instacart, DoorDash, Target, The Knot, Expedia, and Booking.com already have apps live or announced (Grocery Dive). The Agentic Commerce Protocol remains active as the open standard (Apache 2.0) that governs how AI agents, merchants, and payment processors interact (OpenAI Developer Docs).

The OpenAI spokesperson’s own words point in this direction: checkout is “transitioning to apps.” Whether that transition is a full retreat from native checkout or a gradual shift, the implication is the same: OpenAI handles the demand, the merchant handles the conversion.

What this means for AI distribution

If the reports prove accurate, this is the most significant structural signal the AI distribution market has received.

OpenAI, with 800 million weekly active users, would effectively be telling every company: if you want to sell through AI, build your own AI storefront. Not a listing. Not a product feed. An app that you own, with your brand, your compliance, your conversion flow, your data.

For companies in regulated industries (insurance, banking, credit, healthcare), this was always the only viable model. You cannot outsource the transaction to a third-party chat interface when you have regulatory obligations around disclosure, advice boundaries, data handling, and consumer protection. The merchant-owned app is not just a better experience. It is a compliance requirement. This is the architecture WaniWani was designed around from day one: companies own their AI storefront, and the infrastructure underneath handles distribution, analytics, and compliance across every AI platform.

For e-commerce and retail, this is a newer realization. But the economics point in the same direction. Owning your AI storefront means no 4% platform tax, full control over the customer relationship, and the ability to build conversion flows that match your product’s complexity.

Morgan Stanley projects that agentic commerce could represent $190 billion to $385 billion in U.S. e-commerce spending by 2030, capturing 10% to 20% of market share (Morgan Stanley, 2026). The infrastructure decisions being made now will determine who captures that demand and who is invisible.

The window is open, and it won’t stay open

There is a deeper reason why being first matters in AI distribution, and it is not just about early traffic.

AI models learn what tools are available to them. When a user asks ChatGPT about insurance and a quoting app exists, the model learns to reach for it. It learns the app’s capabilities, its response patterns, the types of questions it handles well. Over time, the model develops a preference, not because it was programmed to, but because the app reliably solves user problems. That is how recommendation engines have always worked: the products with the most interaction data win.

When your app is not there, the model learns something too. It learns to recommend whoever is. It learns to route users to competitors. It builds associations between your category and someone else’s brand. Every day you are live, you are training the channel to include you. Every day you are not, you are training it to forget you.

This is why Tuio’s decision to launch the first insurance app on ChatGPT in February 2026, built on WaniWani’s infrastructure, was not a marketing stunt. It was a distribution infrastructure decision. Not a checkout button (insurance does not work that way), but a merchant-owned AI app that handles the full journey: understanding the customer’s context, generating a personalized estimate, explaining coverage, and capturing the lead. The kind of experience that only works when the merchant owns it. WaniWani now powers AI storefronts across insurance and financial services, with over a dozen apps in the approval pipeline across ChatGPT, Claude, and Gemini.

Forrester describes the current moment as “an experimental phase” for agentic commerce (Forrester, March 2026). That is accurate. But experiments have outcomes. The early web was experimental too, and the companies that built their own sites in 1998 did not do it because the economics were proven. They did it because they understood that distribution channels reward presence, and absence compounds.

Whether or not OpenAI officially confirms the end of native checkout, the direction is clear. Discovery happens on the platform. The transaction happens in your storefront. The companies that build one will capture AI-sourced demand. The ones that wait will find, when they finally arrive, that the channel has already learned to work without them.

Frequently Asked Questions

Is OpenAI really killing Instant Checkout in ChatGPT?

OpenAI has not officially confirmed discontinuing Instant Checkout. However, The Information reported in March 2026 that OpenAI is scaling back its shopping plans, and an OpenAI spokesperson reportedly stated that Instant Checkout is “transitioning to apps.” Only about 12 of over one million eligible Shopify merchants had integrated. Forrester’s March 2026 survey confirmed that completing a purchase inside an AI platform is the least-adopted use case among regular users. The direction of travel appears clear, even if the formal announcement has not been made.

What is replacing ChatGPT Instant Checkout?

According to reports, checkout is moving into merchant-owned apps built on top of ChatGPT. Instead of processing transactions natively, ChatGPT would focus on product discovery and research, then route users to apps like Instacart, DoorDash, Target, and Expedia where the merchant owns the transaction experience. The Agentic Commerce Protocol (ACP) remains the open standard governing these interactions.

What is an AI storefront?

An AI storefront is a merchant-owned app built on protocols like MCP that represents a company’s products inside AI conversations. Unlike a product listing or a checkout button, an AI storefront handles the full journey: product discovery, personalization, quoting, compliance, and lead capture. The merchant owns the app, the data, and the customer relationship.

How does AI distribution differ from GEO (Generative Engine Optimization)?

GEO focuses on monitoring and improving how a brand appears in AI-generated responses, addressing top-of-funnel visibility. AI distribution goes further: it enables the full journey from discovery through conversion inside AI platforms, using merchant-owned apps that can answer questions, generate personalized quotes, and capture leads. GEO tells you how AI talks about you. AI distribution lets you sell through AI.

Why does first-mover advantage matter in AI distribution?

AI models learn which tools are available to them through usage. When users ask questions and an app reliably provides good answers, the model develops a preference for routing similar queries to that app. Companies that are live today accumulate interaction data that reinforces their position. Companies that wait will find the channel has learned to work without them.

Sources:

The Information. “OpenAI Scales Back Shopping Plans for ChatGPT.” March 2026. https://www.theinformation.com/articles/openai-scales-back-shopping-plans-chatgpt
Forrester. “What It Means That The Leader In ‘Agentic Commerce’ Just Pulled Back.” March 2026. https://www.forrester.com/blogs/what-it-means-that-the-leader-in-agentic-commerce-just-pulled-back/
PYMNTS. “Shopify Merchants to Pay 4% Fee on ChatGPT Checkout Sales.” January 2026. https://www.pymnts.com/news/ecommerce/2026/shopify-merchants-to-pay-4percent-fee-on-sales-made-through-chatgpt-checkout/
Retail Brew. “Shopify merchants have to pay an additional 4% fee if customers buy via ChatGPT.” January 2026. https://www.retailbrew.com/stories/2026/01/29/shopify-merchants-have-to-pay-an-additional-4-fee-if-customers-buy-via-chatgpt
Marketplace Pulse. “ChatGPT’s 4% Fee Confirms Marketplace Economics.” 2026. https://www.marketplacepulse.com/articles/chatgpts-4-fee-confirms-marketplace-economics
Lengow. “OpenAI’s E-Commerce Bet: What Went Wrong with ChatGPT Checkout.” March 2026. https://blog.lengow.com/chatgpt-wanted-to-become-the-worlds-biggest-shop/
OpenAI. “Introducing shopping research in ChatGPT.” 2026. https://openai.com/index/chatgpt-shopping-research/
OpenAI. “Buy it in ChatGPT: Instant Checkout and the Agentic Commerce Protocol.” 2025. https://openai.com/index/buy-it-in-chatgpt/
OpenAI Developer Docs. “Agentic Commerce Protocol.” https://developers.openai.com/commerce/
Grocery Dive. “Instacart, Uber, DoorDash bringing food delivery to ChatGPT.” 2025. https://www.grocerydive.com/news/chatgpt-instacart-doordash-uber-target-grocery-delivery/802554/
Marketplace Pulse. “Google Kills Its Shopping Marketplace.” https://www.marketplacepulse.com/articles/google-kills-its-shopping-marketplace
Morgan Stanley. “Agentic Commerce Impact Could Reach $385 Billion by 2030.” 2026. https://www.morganstanley.com/insights/articles/agentic-commerce-market-impact-outlook
Instacart. “Instacart App Launches in OpenAI ChatGPT.” 2025. https://investors.instacart.com/news-releases/news-release-details/instacart-app-launches-openai-chatgpt-first-company-offer-new