OpenAI launches ChatGPT Wallet for secure AI agent purchases
The new tool allows US Pro users to let AI agents buy items using a secure payment token system that preserves card rewards.
The short version
- Available to a limited number of US-based ChatGPT Pro users over the age of 18.
- Supports Visa, Mastercard, and American Express cards.
- Uses a payment token system so merchants never see the user's actual card details.
- Requires biometric approval, like Face ID or a fingerprint, for every transaction.
- Allows users to keep existing credit card rewards and dispute protections.
Why it matters
This marks a shift toward 'agentic' commerce where AI can perform complex tasks like shopping autonomously. It attempts to balance the convenience of automation with the security of biometric verification.
A new way to shop with AI
OpenAI has introduced ChatGPT Wallet, a tool designed to let AI agents conduct purchases for users. This feature allows people to link their credit or debit cards to their AI assistant without having to manually enter payment information every time they want to buy something. The goal is to make it easier for agents to complete tasks that involve online transactions.
The service is currently being rolled out to a limited group of ChatGPT Pro users located in the United States. To be eligible for this feature, users must be at least 18 years old and possess a valid US-issued card. The system currently supports major card networks, including Visa, Mastercard, and American Express.
By integrating a wallet directly into the AI experience, OpenAI is moving toward a model of delegated commerce. Instead of a human clicking 'buy' on every single item, the AI can navigate the process of finding and purchasing goods. This transition requires a high level of trust between the user, the AI, and the payment infrastructure.
How the security works
Security is a primary focus for the new wallet system, as OpenAI wants to ensure that card details remain private. When a user adds a card, the system uses a payment token as a substitute for the actual card information. This means that when an agent makes a purchase, the business or the agent itself never sees the user's real card details.
Users still maintain a high level of control over every individual transaction. Before a purchase is finalized, the user must review the specific items, the total cost, and the card being used. After this review, the user must provide a security check, such as a fingerprint scan or Face ID, to authorize the payment. This ensures that the AI cannot spend money without a human giving the final green light.
One of the specific benefits of this setup is that it preserves the user's existing financial perks. Because the transaction is linked to the original card, users can still earn airline credits or other rewards. It also ensures that standard protections remain in place, allowing users to dispute transactions if a product arrives defective or the service is not as described.
The path to full automation
OpenAI envisions a future where users can delegate their shopping habits entirely to their AI agents. In this scenario, users would set specific parameters for the AI, such as spending limits, specific categories of goods, or certain time periods. The AI would then operate within those boundaries to make purchases automatically, only asking for biometric approval when necessary.
To support this vision, OpenAI is also releasing a Commerce Partner Playbook for businesses. This guide explains how companies can integrate their platforms with the new service. It includes technical instructions on how to configure firewalls or content delivery networks so that they do not accidentally block ChatGPT agents from accessing their sites.
The company wants to ensure that the infrastructure of the internet is ready for AI-driven commerce. By providing these guidelines, OpenAI hopes to prevent technical hurdles from stopping agents from completing legitimate purchases. This move signals a push to make AI agents a standard part of the online retail landscape.
Trust and retail hurdles
Despite the technical progress, there are significant concerns regarding trust and security in AI-driven payments. Research from Visa indicates that only 23% of US consumers currently trust generative AI to handle payments on their behalf. In contrast, 61% of those same consumers expressed trust in Visa to handle these types of agentic transactions.
Financial institutions have also raised alarms about the risks associated with this technology. A paper published in September by six major banking groups highlighted the potential for increased rates of scams, fraud, and disputes. One specific concern is that an AI agent could be tricked into purchasing counterfeit goods or falling for fraudulent offers online.
Retailers are also reacting to the rise of AI agents in their stores. For instance, Amazon recently blocked Meta's Muse AI assistant from shopping on a user's behalf. The company stated that the use of an unauthorized AI agent violated its terms and conditions, showing that some platforms may resist the integration of autonomous shopping tools.
What to watch
- Whether more retailers will block AI agents from making purchases.
- How the 'Commerce Partner Playbook' influences how websites handle AI traffic.
- If consumer trust in AI payments increases as more people use the wallet.
Try it
Check your ChatGPT Pro settings to see if the Wallet feature or 'dots' agent tool has been enabled for your account.
Sources and links
- OfficialCommerce Partner Playbook developers.openai.com
- OfficialChatGPT Wallet openai.com
- ReportingPCMag: the article we sourced
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.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by PCMag · Read the original. 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.
