.png){kind=logo}
OpenAI wants ChatGPT to do more than find information and answer questions. The company has introduced ChatGPT Wallet, a new feature that lets its AI agents help users shop online and pay for purchases.
Announced on November 13, the wallet has been developed with payment companies including American Express, Mastercard and Visa. It allows an AI agent to prepare a purchase, while the user remains in charge of approving the payment.
Users can ask ChatGPT to find something they want to buy, and the agent can take care of much of the process. They still get to check the order and payment details before money changes hands.
For now, OpenAI is limiting access to eligible ChatGPT Pro subscribers in the US who are at least 18 years old and have supported US-issued payment cards. The feature works with OpenAI's AI agents, known as dots. Once an agent has selected a product and prepared the order, the user can review the items and the final cost.
Also Read: Why are Nonprofits Suing OpenAI, Google and Other Major AI Companies?
A security check, such as Face ID or fingerprint authentication, is required before the transaction goes through. So, despite the move towards automated shopping, ChatGPT cannot simply spend a user's money without permission.
Sharing card information with an AI system may concern users, particularly when it is handling purchases. According to OpenAI, ChatGPT Wallet addresses this through payment tokens. Instead of giving the AI agent access to actual card details, the system uses tokens to process transactions. This means the original payment information is not directly exposed to the agent or merchant.
Also Read: OpenAI Introduces Red Team and Defense Factory to Accelerate Enterprise Security
OpenAI also sees room to expand the feature. Users could eventually set spending limits, choose which types of products an agent can buy and decide when it can make purchases. For now, every transaction needs individual approval. But the longer-term plan suggests OpenAI wants ChatGPT to handle more of the online shopping process, rather than simply recommending what users should buy.