ChatGPT Wallet Launches: OpenAI Dots Agents Can Now Place Orders for You

OpenAI launches ChatGPT Wallet for Dots agents, enabling AI to complete payments on users' behalf.
OpenAI has launched ChatGPT Wallet for its Dots AI agent platform, enabling agents to complete purchases within authorized limits and closing the execution loop from price comparison to payment. The initial rollout is strictly limited to US ChatGPT Pro subscribers aged 18+ with a US billing address and eligible US-issued Amex, Chase, Mastercard, or Visa cards. The article notes OpenAI chose to start from user payment credentials rather than deep merchant integration, and that the core challenge lies in authorization boundaries, risk control, refunds, and liability.
OpenAI is moving AI agents from "talking" to "doing." According to RSS feed reports, after launching the Dots AI agent platform earlier this year, OpenAI has now rolled out a companion wallet for these agents—ChatGPT Wallet. Starting today, the feature is gradually becoming available to ChatGPT Pro subscribers in the US who are 18 or older, use a US billing address, and hold eligible US-issued American Express, Chase, Mastercard, and Visa cards. Once a user adds a bank card, agents can complete purchases on their behalf within authorized limits.

From Conversation to Transaction: A Key Step for Dots Agents
Dots is the AI agent platform OpenAI launched earlier this year, positioned to let models do more than answer questions—they can execute multi-step tasks on behalf of users. The launch of ChatGPT Wallet fills in the "payment execution" link for agents. Previously, agents could help you compare prices, filter products, and fill out forms, but a human still had to take over at the final payment step; the wallet feature closes this gap.
This means OpenAI has chosen a pragmatic direction for agent commercialization: rather than deeply coupling with the merchant side, it starts from the user's payment credentials and lets agents complete the transaction loop as a "proxy consumer." For users, the benefit is skipping the last few steps; for OpenAI, it opens the possibility of upgrading agents from information-processing tools to transaction entry points.
AI Agent refers to an AI system that can perceive its environment, plan autonomously, and invoke tools to execute multi-step tasks to achieve goals, distinct from chatbots that only handle single-turn Q&A. Its typical capabilities include task decomposition, calling external APIs, reading and writing files, browsing the web, and executing transactions. The industry often uses "autonomy" and "tool use" as the dividing line for agent maturity: only when an agent can autonomously complete the full chain of "price comparison—filtering—form filling—payment" is the execution loop truly closed. OpenAI's Dots platform is designed along these lines, upgrading the model from "giving suggestions" to "acting on your behalf," and enabling the payment step is the highest-risk and most commercially valuable part.
Availability and Eligibility Requirements
According to reports, ChatGPT Wallet's initial rollout has clear restrictions:
- US users only, and must be at least 18 years old;
- Must use a US billing address;
- Only supports US-issued American Express, Chase, Mastercard, and Visa cards;
- Available only to ChatGPT Pro subscribers.
These restrictions show that OpenAI has adopted a conservative strategy for the high-risk payment step: first prove the process with users who have high willingness to pay and verifiable identities, then consider expansion. The card network whitelist approach also facilitates risk control and liability attribution.
Card network whitelist means a payment product only accepts cards from specific card networks (such as Amex, Visa, Mastercard) and specific issuing banks. It serves three purposes: first, it leverages the card networks' mature risk control and dispute resolution systems to share fraud risk; second, it makes it easier to determine liability in cases of chargebacks or disputes; third, it reduces dependence on merchant-side modifications. Limiting to "US-issued" and "US billing address" is used to meet KYC (Know Your Customer) and anti-money laundering compliance requirements and to avoid the complexity of cross-border payments. This approach is common in the cold-start phase of payment products and is usually gradually relaxed as risk control models mature.
The Trust Problem Behind Agent Payments
Letting AI agents spend money directly—the core difficulty is not technology but the boundaries of trust. Users need to know clearly under what conditions an agent can place orders, what the per-transaction and cumulative limits are, and how to cancel and assign liability if something goes wrong. The original source material does not disclose these details, but from a product design logic perspective, the wallet feature must come with authorization mechanisms and transaction records.
Another dimension worth watching is the merchant side. When a transaction is initiated by an agent, how merchant risk control systems identify it and whether it is treated as normal consumer behavior—there is currently no public information. This determines whether agent payments can move from small-scale pilots to everyday use.
Authorization mechanisms in agent payments typically use two approaches: "tokenization" and "limit-based authorization." Tokenization means replacing the real card number with a one-time or restricted payment token, so the agent can invoke it within authorized limits without exposing full card information; limit-based authorization specifies per-transaction amounts, cumulative amounts, validity periods, and permitted merchant categories. A mature agent payment system also requires revocable authorization, transaction detail receipts, freezing of abnormal transactions, and refund dispute handling processes. These mechanisms determine whether users dare to hand their payment credentials to AI, and they are also a regulatory focus—for example, the Strong Customer Authentication (SCA) requirement in the EU's PSD2 imposes additional verification requirements on transactions "not initiated by the cardholder."
Impact on the AI Agent Landscape
ChatGPT Wallet pushes "AI proxy consumption" from concept to a usable product form. For similar agent products, the opening of payment capabilities means the competitive dimension extends from model capabilities to transaction pipelines and trust infrastructure. Whoever can build solid authorization, risk control, refunds, and dispute handling is more likely to make users comfortable handing their wallets to AI.
In the short term, the impact of this feature is limited to US Pro users, but the signal it sends is clear: the next stop for agents is completing real-world operations for users, not just giving suggestions.
One-line Summary
OpenAI launches ChatGPT Wallet for Dots agents, enabling AI to complete payments on users' behalf.
Paragraph Summary
OpenAI has launched ChatGPT Wallet for the Dots AI agent platform introduced earlier this year, enabling agents to complete purchases on users' behalf within authorized limits, closing the execution loop from price comparison and form filling to payment. The initial rollout is strictly limited: gradually available to ChatGPT Pro subscribers in the US who are 18 or older, use a US billing address, and hold eligible US-issued Amex, Chase, Mastercard, or Visa cards. The article notes that this means OpenAI has chosen to start from user payment credentials rather than deep merchant integration, letting agents complete transactions as a "proxy consumer." The core challenge of agent payments is not technology but authorization boundaries, risk control, refunds, and liability mechanisms, and there is still no public information on how merchants will identify agent-initiated transactions.
Related articles

Gemini 4 Neon Review: Benchmark Scores Rival Carbon, API Pricing and Quota Breakdown
Google Gemini 4 Neon is now live, benchmark scores closely tracking Carbon, API pricing 3/15 at twice Hydrogen's rate. This article compiles Bilibili creator testing: Neon can replace 90% of Carbon scenarios, AntiGravity quota consumption is only one-third of Carbon, plus subscription channel restrictions and 2API ban risks.

Gemini 4 Helium Coming Soon? Signals of Google Loosening Subscription Restrictions
Google team member hints releases will continue; community speculates Gemini 4 Helium (Flash Lite) may be next. Subscription ban controversy gets a response as product lead says "Soon," suggesting restrictions may be lifted. This article covers Gemini developments, OpenAI Coding Agent and risk management, Doubao real-time audio-video API, and compute business dynamics.

AI-Assisted React Development in Practice: Optimizing a Video App and Handling Errors
A 14-year veteran developer demonstrates AI-assisted React video app optimization: compressing video from 80MB to 5MB (~94%) via ImageKit, plus reliable error handling and retry mechanisms. Includes planning, code understanding, and verification methods for working with AI.