OpenAI is no longer content to be the engine behind the conversation. It wants to be the register behind the transaction.
The artificial intelligence company best known for ChatGPT has quietly introduced a native shopping feature that lets users browse, compare, and now purchase products without ever leaving the chat interface. The twist: OpenAI plans to charge merchants a 4% fee on every transaction processed through its checkout system. No advertising. No auction-based placement. Just a flat cut of the sale, applied uniformly whether you’re selling a $15 phone case or a $2,000 laptop.
That number — 4% — sounds modest in isolation. It isn’t.
As TechRadar reported, this fee structure positions OpenAI not as a search engine or an advertising platform but as a commerce intermediary, a toll collector sitting between consumer intent and merchant fulfillment. The implications for retailers, payment processors, and the broader digital economy are significant and, in some cases, deeply uncomfortable.
To understand why, consider the economics. Credit card processing typically costs merchants between 1.5% and 3.5%, depending on the card network, the processor, and the merchant’s volume. OpenAI’s 4% fee would come on top of that. So a merchant selling through ChatGPT’s checkout could face combined transaction costs of 5.5% to 7.5% per sale. For businesses operating on thin margins — electronics retailers, grocery delivery services, commodity goods sellers — that’s a punishing toll. For higher-margin categories like fashion or beauty, it’s more digestible, but still material.
And OpenAI isn’t offering to absorb the payment processing costs. The 4% is its fee alone.
The company has framed this as a cleaner alternative to the ad-driven model that dominates platforms like Google Shopping and Amazon. There are no sponsored listings. No pay-to-play rankings. Product recommendations within ChatGPT are ostensibly determined by relevance, user preferences, and AI-driven matching rather than by who bid the most for placement. OpenAI has emphasized that its shopping results are not influenced by advertising relationships, a claim that, if sustained, would represent a genuine departure from how product discovery works on the modern internet.
But merchants aren’t naive. A 4% fee on every checkout is, functionally, a tax on access to ChatGPT’s growing user base. And that user base is enormous. OpenAI reported earlier this year that ChatGPT has surpassed 400 million weekly active users, a figure that dwarfs many standalone e-commerce platforms. If even a fraction of those users begin making purchases through the chat interface, the volume could be staggering.
The shopping feature itself has been rolling out incrementally. Users can ask ChatGPT for product recommendations — “best running shoes under $150” or “wireless earbuds for commuting” — and receive curated results with images, prices, reviews, and direct purchase links. The native checkout, which keeps the entire transaction within ChatGPT rather than redirecting to a merchant’s website, is the newest and most consequential addition. It transforms ChatGPT from a referral source into a point of sale.
This matters because it changes who owns the customer relationship. When a shopper clicks through from Google to a retailer’s site, the retailer captures that customer’s data, email address, browsing behavior, and purchase history. When the transaction happens inside ChatGPT, OpenAI sits in the middle. The merchant gets the order. OpenAI gets the data. And the customer may never visit the merchant’s own storefront at all.
For small and mid-sized retailers, this creates a genuine dilemma. Participating in ChatGPT’s commerce system could drive meaningful incremental sales, especially as consumers increasingly use AI assistants for product research. But it also means surrendering margin and customer data to a platform that could, over time, become as dominant and extractive as the ones they’re already struggling with.
Amazon takes roughly 15% on most third-party sales, plus fulfillment fees. Shopify charges transaction fees ranging from 0.5% to 2% on top of monthly subscription costs. Apple and Google take 15% to 30% on app-based purchases. In that context, 4% looks almost generous. Almost.
The concern isn’t today’s rate. It’s tomorrow’s.
Platform economics follow a familiar pattern: attract participants with favorable terms, build dependency, then extract. Amazon’s marketplace fees have risen steadily over the past decade. Apple’s App Store commission was 30% from the start, and it took antitrust pressure and developer revolt to carve out a 15% tier for smaller developers. OpenAI’s 4% could be an introductory rate, a loss leader designed to onboard merchants before the real pricing begins.
OpenAI has not publicly committed to keeping the fee at 4% indefinitely. As TechRadar noted, the company’s broader business model is still evolving rapidly, with the firm burning through cash to fund compute infrastructure while simultaneously exploring multiple revenue streams including subscriptions, API access, and enterprise licensing. Commerce fees could become a critical pillar of that revenue mix, particularly if the subscription model faces growth headwinds.
The timing of this move coincides with a broader industry shift toward AI-mediated commerce. Google has been integrating shopping features into its AI Overviews and Gemini assistant. Amazon is embedding AI more deeply into its product search and recommendation systems. Perplexity, the AI search startup, has also experimented with commerce integrations. The race to become the default AI shopping assistant is accelerating, and the stakes are high because whoever controls the point of purchase controls the most valuable moment in the consumer funnel.
Sam Altman and OpenAI appear to be betting that consumers will value the convenience of buying through a conversational interface enough to make it a habit. The pitch is straightforward: instead of opening multiple browser tabs, comparing prices across sites, reading reviews on yet another platform, and then navigating a checkout flow riddled with upsells and dark patterns, you just tell ChatGPT what you want and buy it. One conversation. One transaction.
It’s compelling. It’s also a concentration of commercial power that would have been unimaginable five years ago.
Payment industry analysts have raised questions about how OpenAI’s checkout system will handle disputes, returns, and fraud — the messy operational realities that make commerce difficult at scale. When a customer buys through ChatGPT and the product arrives damaged, who do they contact? When a merchant suspects a fraudulent transaction processed through OpenAI’s system, what recourse exists? These aren’t hypothetical concerns. They’re the daily reality of e-commerce operations, and OpenAI has limited experience managing them.
Stripe, which has a deep partnership with OpenAI and powers much of its payment infrastructure, will likely handle the technical plumbing. But the customer-facing responsibility — the brand trust that makes someone comfortable entering a credit card number — rests with OpenAI. That’s a different kind of trust than making a chatbot that writes decent poetry.
There’s also a competitive dimension that deserves scrutiny. If OpenAI’s AI determines which products to recommend and OpenAI collects a fee on purchases, there’s an inherent incentive to recommend products that are more likely to convert — that is, products that generate fees for OpenAI. The company says its recommendations aren’t influenced by commercial relationships. But when the recommender and the toll collector are the same entity, maintaining that wall requires extraordinary discipline and transparency. Neither has been a hallmark of the AI industry so far.
Regulators will eventually take notice. The European Union’s Digital Markets Act already imposes obligations on platforms that serve as gatekeepers to commerce. If ChatGPT becomes a significant channel for product sales in Europe, it could trigger regulatory scrutiny under the DMA’s provisions regarding self-preferencing and fair access. In the United States, the FTC has been increasingly attentive to the ways AI systems can embed commercial bias into ostensibly neutral recommendations.
But regulation moves slowly. Markets don’t.
For now, merchants face a practical calculation. The cost of customer acquisition through traditional digital advertising — Google Ads, Meta ads, TikTok campaigns — has been rising for years. The average cost-per-click in competitive e-commerce categories can exceed $2 to $5, and conversion rates hover around 2% to 3%. Do the math on that and the effective cost of acquiring a sale through paid advertising often exceeds 10% to 15% of the transaction value. Compared to that, a flat 4% fee with no upfront advertising spend looks attractive, at least on a per-transaction basis.
So some merchants will sign up eagerly. Others will resist. And the ones who sign up eagerly may find, three or four years from now, that they’ve built a dependency on a channel they don’t control, selling to customers they can’t reach independently, at a fee rate that has quietly crept upward.
This is the pattern. It played out with Amazon Marketplace. It played out with the App Store. It played out with Google Shopping. The specific mechanics differ, but the structural dynamic is the same: a platform with massive consumer reach offers merchants access in exchange for a fee, the fee starts reasonable, and then the platform’s market power allows it to ratchet up the cost because by then the merchants can’t afford to leave.
OpenAI’s 4% checkout fee isn’t just a pricing decision. It’s a declaration of intent. The company that built the most popular AI assistant in history now wants to be the place where you buy things, and it wants a piece of every sale. Whether that ends up being good for consumers, good for merchants, or merely good for OpenAI is a question that won’t be answered by a chatbot. It’ll be answered by the market — and, eventually, by the regulators trying to keep up with it.
WebProNews is an iEntry Publication