Skip to content
17 November 2026

ChatGPT’s new AI shopping wallet rolls out to US Pro users

OpenAI’s new ChatGPT Wallet gives AI agents the power to shop for you, but it also raises new security questions.

ChatGPT’s new AI shopping wallet rolls out to US Pro users

On Friday, OpenAI added a ChatGPT Wallet to its suite of AI tools, allowing its always-on agents, called dots, to act as personal shoppers. The feature is being offered first to a limited pool of United States-based ChatGPT Pro subscribers who are at least 18 years old and who own a Visa, Mastercard or American Express credit card issued in the U.S.

How the ChatGPT Wallet transforms AI-powered shopping

Users start by entering their payment details a single time through a secure web form. The information never travels to the AI itself; instead, the card network returns a payment token that contains a surrogate card number, security code and expiry date. When a dot proposes a purchase, the token is supplied to the merchant, while the real card data stays private.

Biometric confirmation before checkout

Before any transaction is finalized, the dot displays the selected items, the total amount and the masked token-based card to the user. Approval must be granted via Face ID, Touch ID or a fingerprint scan. This step creates an immutable log of the user’s consent, ensuring that the AI cannot complete a purchase without explicit, biometric verification.

Token-based payments keep card details hidden

The payment token acts as a stand-in for the real card, meaning merchants and the AI never see the actual account number. Users retain all standard credit-card benefits, such as rewards points and dispute rights, because the transaction is still anchored to the original issuing bank.

Rollout details and what’s coming next

At launch, the Wallet is limited to U.S. ChatGPT Pro accounts with eligible credit cards. Debit cards are not supported, and the feature works only when a conversation involves a dot. OpenAI plans to broaden retailer compatibility, but some large e-commerce sites, most notably Amazon, have explicitly blocked AI-driven checkout attempts to protect their own voice-assistant ecosystems.

Future updates aim to give users finer control: spending caps, pre-approved categories, and time-based limits could allow a dot to finalize purchases without a per-transaction biometric prompt. OpenAI has also released a “Commerce Partner Playbook” to help merchants adjust fraud filters and accept signed agent requests.

Privacy, security and user safeguards

While the convenience is evident, critics warn that handing an AI access to financial data introduces new risk vectors. James E. Lee, president of the Identity Theft Resource Center, describes the technology as “frontier” and advises only risk-tolerant early adopters to experiment. He stresses the need to understand data storage, permission revocation and dispute procedures before enabling an agent to act on one’s behalf.

Surveys show that a growing share of Americans already share personal details with generative AI tools: a 2024 Cisco study found that 24-29 % of respondents have entered financial information, while 37 % disclosed health data. Some providers may retain conversation snippets for model training, making it essential for users to read terms of service and opt out of data-use programs where possible.

Practical steps to protect oneself include: disabling model-training data collection, avoiding uploads of documents that contain Social Security numbers or bank details, using temporary or incognito chat windows when discussing sensitive topics, and regularly reviewing privacy settings. For recurring subscriptions, users should monitor merchant-issued token usage, as the stored token can be charged again without further biometric approval.

Overall, the ChatGPT Wallet showcases how AI can streamline everyday transactions, yet it also serves as a reminder that convenience must be balanced with vigilant privacy management.

Author

Ethan Carver

Ethan Carver writes about Deep Tech for readers of this website, unpacking complex innovations, assessing practical impacts, and translating advanced research into clear, actionable insight. He covers breakthroughs, platform architectures, AI infrastructure, and hardware-software convergence with an analytical, accessible approach and draws on extensive experience in the sector to clarify what emerging technologies mean for technologists and decision makers.