AI shopping vision where ChatGPT gets a direct checkout button, performing the role of an agentic gateway, faces merchant, consumer, and technical headwinds
OpenAI appears to be quietly scaling back its earlier ambition to enable direct product checkout inside ChatGPT, signaling a notable shift in how the company approaches AI-driven commerce.
According to recent reporting citing internal discussions and sources familiar with the project, OpenAI has abandoned or significantly reduced plans to process payments directly within the chatbot, opting instead to route purchases through third-party apps and retailer platforms.
Notably, OpenAI has not published an official announcement or detailed explanation regarding the change. The pivot has instead emerged through media reports and industry analysis, leaving observers to piece together the likely reasons behind the strategic recalibration.
The shift marks a contrast with OpenAI’s messaging in 2025, when the company introduced Instant Checkout inside ChatGPT as part of its broader push toward “agentic commerce” — a model in which AI assistants could recommend products and proceed to completing purchases on behalf of users. This step is key to potentially transforming ChatGPT into a consumer interface that could compete with traditional apps and marketplaces.
However, the latest reports suggest a more modest direction. Instead of processing payments directly, ChatGPT is expected to function primarily as a product discovery and recommendation layer, directing users to retailer websites or apps where transactions are finalized.
Low merchant adoption appears to be a key factor in a scale-down
One of the most immediate obstacles appears to have been limited merchant participation. Despite the theoretical reach of ChatGPT’s hundreds of millions of users, only a small number of merchants integrated the checkout system, according to reporting based on internal data and partner feedback.
In some cases, the complexity of onboarding, including requirements around inventory synchronization, shipping integration, and payment processing, made the system difficult to implement at scale. For large retailers already operating sophisticated commerce platforms, the additional integration layer may have offered limited incremental benefit compared with directing customers to existing storefronts.
Another possible reason for low adoption among sellers could be simple scepticism caused by recent failed agentic commerce initiatives, like Amazon backlash when small retailers discovered their products listed and sold on Amazon without their consent, raising a firestorm of concerns about transparency, consent, and seller rights, as well as concerns about potential surveillance brought up by OpenAI’s recent Pentagon deal.
Consumer behavior: browsing more than buying
Another challenge may have been the mismatch between AI discovery and actual purchase behavior. Reports indicate that many users engaged ChatGPT to research products, compare options, or generate buying guides, but ultimately completed purchases on external platforms where they already had accounts, saved payment methods, and loyalty programs.
This behavior aligns with broader consumer research on AI shopping tools. A 2025 survey found that only 34% of U.S. consumers were comfortable allowing AI assistants to make purchases on their behalf, highlighting lingering trust barriers around automated transactions.
Meanwhile, 59% of consumers say they are open to using chatbots for faster service, suggesting that conversational AI may currently be better suited to customer support and product discovery rather than final payment execution.
Technical and compliance challenges
Beyond adoption issues, direct checkout inside a conversational interface introduces significant operational and regulatory complexity.
Running a full e-commerce stack requires systems for payments, refunds, fraud prevention, taxation, shipping logistics, and inventory updates. According to industry reporting, these challenges proved difficult to scale across thousands or millions of merchants within a chatbot environment.
Fraud management alone can become particularly complicated when purchases are executed by AI agents or through conversational flows, raising questions about liability, identity verification, and consumer protection.
Chris Jones, Managing Director at PSE Consulting, has shared his perspective on why the scale-back reflects broader structural challenges across commerce systems:
“The news that OpenAI has scaled back Instant Checkout should not be seen as a failure of agentic commerce. Instead, it reflects the ambition of integrating autonomous AI agents into real world shopping at scale. The recent Adyen report on agentic commerce highlights the same structural challenges. Based on insights from more than 200 enterprise merchants and direct work with AI commerce platforms, it identifies five constraints that go beyond any one company’s control: protocol fragmentation, product data not designed for machine consumption, legacy enterprise stacks, trust and liability frameworks, and onboarding at scale. Together these factors mean that what works in demos can break quickly in production environments. From a payments perspective, some challenges remain addressable. Fraud, tax and payment flows can be improved through collaboration across payment networks, infrastructure providers and AI platforms. These are areas where scale, standards and partnerships can drive progress over time.”
Strategic implications for AI commerce
The retreat from in-chat checkout does not necessarily signal the end of AI-driven shopping. Instead, it may reflect a more realistic near-term architecture for AI commerce, where conversational assistants act as discovery engines and orchestration layers rather than full transactional platforms, while integrating with existing digital commerce infrastructure.
In that sense, OpenAI’s shift may be treated as an early correction in the still-emerging market for AI-mediated commerce interfaces. For now, however, even for companies at the forefront of generative AI, turning conversational intelligence into a fully functional commerce platform remains far more complex than generating product recommendations in a chat window.
The article was updated on March 16, 2026 with expert commentary by Chris Jones, Managing Director at PSE Consulting.
