OpenAI is shifting its strategy for commerce within ChatGPT, moving away from plans to allow direct purchases within the chatbot interface. Instead, the company will redirect users to third-party apps and websites to complete transactions, a change that reflects both user behavior and significant technical hurdles. This marks a retreat from the ambitious “instant checkout” feature launched last year, which aimed to streamline online shopping directly within the AI-powered platform.
The move, first reported by The Information, signals a recalibration of OpenAI’s approach to integrating shopping into its popular chatbot. An OpenAI spokesperson stated that the company is “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.” The decision underscores the complexities of building a fully functional e-commerce ecosystem within an AI assistant.
Launched in September, OpenAI’s Instant Checkout partnered with major commerce platforms like Shopify and Etsy, bringing millions of product listings into ChatGPT. The system relied on the Agentic Commerce Protocol, developed with Stripe, to create a framework for AI agents, merchants, and payment systems. Still, early data revealed a key disconnect: while users were actively using ChatGPT for product discovery and comparison, very few were actually completing purchases within the chatbot itself.
Harley Finkelstein, president of Shopify, recently noted that only about a dozen merchants on the platform—despite millions of total users—are currently utilizing AI tools for sales. He attributed this leisurely adoption to the ongoing development of the necessary AI infrastructure, stating that companies are waiting for AI agents and commerce integrations to mature.
The Challenge of Seamless Integration
Beyond user preferences, significant technical and regulatory challenges contributed to the shift. Integrating ChatGPT directly with merchant systems requires real-time synchronization of product data—price, availability, shipping details, and inventory—across potentially millions of retailers. Maintaining this level of accuracy is a formidable task, as even brief discrepancies can lead to failed transactions or incorrect orders.
Modern e-commerce platforms often update inventory information in intervals of just a few minutes to ensure accuracy. However, a chatbot interface demands instant synchronization. If a user asks ChatGPT to purchase an item that is suddenly out of stock, the system must detect that change *before* processing the payment. This becomes exponentially more difficult when dealing with a vast and diverse network of merchants operating on different software and systems.
Security, Compliance, and the Cost of Building an E-Commerce Platform
Security and fraud prevention also played a crucial role in the decision. According to Juniper Research, global e-commerce losses due to fraud reached an estimated $48 billion in 2023. Integrating AI agents into the payment process introduces recent vulnerabilities and requires robust fraud detection systems to verify the legitimacy of purchases initiated by the chatbot.
Compliance with tax regulations and consumer protection laws presented another hurdle. In the United States, for example, companies must navigate a complex web of state and local sales tax jurisdictions. As of February, OpenAI had not yet developed a system to automatically calculate and remit these taxes, a standard feature of established e-commerce platforms like Amazon and eBay.
Essentially, building a fully functional checkout platform within ChatGPT would require replicating much of the infrastructure that existing e-commerce companies have spent years developing and refining. Without these systems in place, scaling direct purchases within the chatbot proved unsustainable.
The Future of AI-Powered Commerce
Despite stepping back from direct checkout, OpenAI remains committed to developing commerce infrastructure. The company and Stripe will continue to collaborate on the Agentic Commerce Protocol, aiming to define standards for interactions between merchants, payment providers, and AI systems. This ongoing work suggests a long-term vision for AI-driven commerce, even if the initial approach proved too ambitious.
OpenAI’s shift also casts a spotlight on Google’s Universal Commerce Protocol (UCP), an open standard designed to support commerce interactions across various platforms. Unlike OpenAI’s initial approach, UCP separates the product discovery and transaction stages, allowing AI systems to surface product information while leaving the actual purchase to the merchant’s own checkout system. Google’s approach, as outlined in Google’s UCP documentation, relies on merchants providing structured product data that AI systems can access without directly handling payment processing or inventory management.
This strategy aligns with Google’s existing infrastructure, which already supports product discovery through Google Shopping and its Merchant Center, connecting millions of retailers and providing structured data on pricing and availability. The move by OpenAI appears to be converging with this model, prioritizing product discovery and referral while leaving the complexities of checkout to established e-commerce platforms.
Looking ahead, OpenAI will continue to refine its commerce strategy within ChatGPT, focusing on facilitating product discovery and directing users to trusted retailers for purchase. The company’s ongoing collaboration with Stripe on the Agentic Commerce Protocol suggests a continued commitment to shaping the future of AI-powered shopping, albeit through a more measured and collaborative approach.
What are your thoughts on OpenAI’s decision? Share your comments below and let us know how you apply AI for shopping.
