OpenAI adds AI payment capabilities to ChatGPT with major card networks
OpenAI is partnering with Visa, Mastercard, and Amex to let AI agents make purchases for users using biometric security.
The short version
- OpenAI partnered with Visa, Mastercard, and American Express to enable AI-driven commerce.
- Users authorize transactions using biometric verification like facial recognition or fingerprints.
- The service is currently rolling out to ChatGPT Pro subscribers in the United States.
- Tokenized credentials ensure that actual card details are never shared with merchants or AI agents.
- OpenAI plans to eventually allow purchases within preset spending limits without per-transaction approval.
Why it matters
This marks a shift from AI simply recommending products to actually executing transactions. It paves the way for autonomous commerce where AI handles the logistics of online shopping.
A new era for AI commerce
OpenAI has officially introduced a feature that allows AI agents to perform purchases on behalf of users. This new capability is integrated directly into ChatGPT and works through established payment networks. The company collaborated with three major players to make this possible: Visa, Mastercard, and American Express.
This move signals a transition in how people interact with digital services and online merchants. Instead of just getting a list of product recommendations, users can now have the AI handle the checkout process. It represents a significant step toward a more automated way of conducting business online.
The rollout is currently limited to a specific group of users to ensure a controlled launch. Specifically, the feature is being released to ChatGPT Pro subscribers located in the United States. These users must also hold eligible cards issued within the country to access the new payment functionality.
How the payment system works
Security is a primary focus for the new system, which relies on biometric verification to authorize spending. Users can approve a transaction using tools like fingerprints or facial recognition. This ensures that the AI cannot spend money without a clear, physical signal of consent from the human owner.
To protect sensitive information, the payment networks generate what are known as tokenized card credentials. These are unique digital identifiers created for each specific approved purchase. Because of this process, the user's actual card details are never shared with the AI agents or the merchants involved in the sale.
OpenAI is also working closely with various payment processors and fraud prevention firms to ensure these transactions are safe. They are integrating these new capabilities into existing checkout systems. This allows the AI to navigate standard online shopping flows while maintaining a high level of security for the consumer.
The competitive landscape
OpenAI is not the only company looking at the infrastructure for AI-driven spending. Google entered this space in 2025 by introducing its own Agent Payments Protocol. Additionally, both Visa and Mastercard have launched their own initiatives to support transactions initiated by AI agents.
The cryptocurrency sector is also developing specialized infrastructure for autonomous payments. For example, Coinbase introduced the x402 protocol in 2025 to allow for stablecoin payments for digital services. Meanwhile, Circle launched the Agent Stack in May, which lets agents hold and spend USDC within specific limits.
Other companies are exploring different ways to bridge the gap between traditional finance and AI. Crossmint, a provider of stablecoin infrastructure, has taken a different path by combining stablecoin wallets with traditional card networks. These diverse approaches show that the industry is racing to build the plumbing for an AI-driven economy.
Future spending and limits
While the current system requires approval for transactions, OpenAI has shared its long-term vision for the feature. The company plans to eventually support purchases that occur within preset spending limits. This would allow the AI to make certain types of purchases without needing to ask the user for permission every single time.
These restrictions are intended to give users more control over how much an agent can spend. By setting a ceiling or specific categories for spending, users can delegate routine tasks to the AI. This balance of autonomy and oversight is a key component of the company's strategy for agentic commerce.
The goal is to make the experience as seamless as possible for the end user. As the technology matures, the friction of manual approval is expected to decrease. This will allow AI agents to become more useful for complex, multi-step tasks that involve buying goods or services.
Market growth and projections
The financial scale of this new technology is expected to be massive over the next few years. According to a report from Juniper Research published in April, the value of global agentic commerce transactions is projected to reach $1.5 trillion by 2030. This is a significant jump from the $8 billion estimated for 2026.
This growth suggests that AI agents will become a standard part of the global economy. As more companies adopt these payment systems, the way consumers interact with online merchants will fundamentally change. The shift from human-led clicking to AI-led purchasing is one of the most significant trends in modern tech.
For the reader, this means that the role of the AI will expand from a simple chatbot to a functional assistant. These agents will eventually be able to navigate the web, find products, and complete the final step of the transaction. This evolution will redefine personal productivity and online shopping habits.
What to watch
- How many users will opt into preset spending limits versus per-transaction approval?
- Which other major payment networks will join the integration beyond Visa, Mastercard, and Amex?
Try it
Check if you have a ChatGPT Pro subscription in the US to see if the new payment features are active in your account.
Sources and links
- ReportingTradingView: the article we sourced
First reported by TradingView. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.