OpenAI launches ChatGPT Wallet for purchases through AI agents
publishers 49articles 59first reported 10 Nov, 18:00 UTCdeveloping since 11 Nov · 6 editions
OpenAI launched ChatGPT Wallet on Friday, allowing its Dots AI agents to complete purchases on users’ behalf while keeping their card details hidden from both the agent and the merchant.
The feature is initially rolling out to eligible ChatGPT Pro subscribers in the United States who are at least 18 and have supported U.S.-issued cards. OpenAI developed the wallet with payment networks including American Express, Mastercard and Visa.
Users add an eligible card once, then ask a Dot to help with a purchase. Before checkout, they can review the items, total and selected payment card. Every transaction requires a security check, such as Face ID or fingerprint verification.
At payment, the network substitutes a token for the user’s actual card information. The merchant receives necessary order details, such as the shipping address and contact information, but neither the merchant nor the AI agent sees the card number. OpenAI says users retain their card rewards and standard protections, including the ability to dispute transactions.
The company is limiting the initial rollout while it tests and improves the service. OpenAI eventually plans to let users pre-authorize purchases within restrictions such as spending limits, product categories or specified time periods. Availability may also depend on whether retailers permit AI agents to access their websites and complete transactions.
HOW THIS STORY WAS MADE
Written from 8 articles, 4 with the publisher's own text; 7 independent newsrooms once syndicated copies count as one; this version written 90 h after the record first saw the story; the editor kept it as written.
- Publishers
- 49
- Source articles
- 59
- Given to the writer
- 8, 4 with the article's own text
- Independent newsrooms
- 7
- This version written
- 90 h after the record first saw the story
- Second model (editor)
- kept as written
- Publication gate
- passed
- Human review
- none
WHAT THE EDITOR DID
The second model checked the writer's draft against the same sources and published it as written.
Written by a language model from the sources above, then checked by a second model that may only cut, attribute or correct. How it works →