Skip to content
highrisk.news
RUJoin

News

OpenAI Launches ChatGPT Wallet for AI Agent Purchases With Card Tokenisation

OpenAI has launched ChatGPT Wallet, letting AI agents make purchases without access to card details. It is initially available to ChatGPT Pro subscribers in the US.

HighRisk Media Newsroom, , 2 min read · Читать по-русски

OpenAI Launches ChatGPT Wallet for AI Agent Purchases With Card Tokenisation

OpenAI has launched ChatGPT Wallet, a payment feature that allows AI agents to make purchases on behalf of a user without access to the user's card details. The wallet was developed with American Express, Mastercard and Visa. It will first be available to ChatGPT Pro users in the US. According to Payments Industry Intelligence, the launch reflects the company's move from product discovery to payment execution, which affects merchants, payment providers and anti-fraud systems.

How the wallet works

A user registers an eligible payment card once. After that, their AI agent, which OpenAI has named dots and which runs continuously, can initiate a purchase. Before payment, the user sees the order details, the total amount and the selected card, and confirms the payment with biometrics: facial recognition or a fingerprint. After approval, the payment network issues a token with substitute payment credentials, so neither the agent nor the merchant sees the original card number. Transactions run through the existing card infrastructure, which preserves rewards, chargeback protections and the relationships between consumer, bank and merchant.

Plans for autonomous purchases

OpenAI is also considering a more independent model, in which users set spending limits, merchant categories and timeframes in advance, and the agent buys within those boundaries. For now, the launch is limited to Pro subscribers in one region.

Barriers for merchants and payment providers

Tokenisation addresses data security but not transaction acceptance. Existing anti-fraud systems may treat automated purchases as suspicious, decline legitimate payments or block AI agents from checkout. To help businesses adapt their websites, security controls and payment infrastructure, OpenAI has released a Commerce Partner Playbook. The company is working with payment providers Adyen, Checkout.com and Stripe, and anti-fraud specialists Forter, Riskified, Sardine and Signifyd, to improve recognition of such transactions and raise their approval rates.

User trust

Trust remains the main barrier. According to Visa research, only 23% of Americans trust AI to make payments on their behalf. Separate testing of AI models in realistic shopping scenarios found failures in 43% of runs. In the publication's view, robust authentication, transparent authorisation and consumer protections will remain essential as agentic commerce develops.

Why it matters

  • The emergence of AI agents as an independent transaction participant changes requirements for payment infrastructure and anti-fraud.
  • Tokenisation and keeping the card infrastructure preserve rewards and chargeback mechanisms.
  • Low user trust and model failures in tests could slow mass adoption of such purchases.

Sources

Prepared by AI from the sources above and checked against them (how we work). Spotted a mistake? Write to @highrisk_team.

Get stories like this first

  • 📰 Daily briefing — Telegram channel @highrisk_media
  • 📡 PSP and payment-method alerts — in our bot
  • 🔒 HighRisk Private — closed chat with topics for iGaming, payments, crypto, dating, nutra
  • 📄 Post your resume — seen by industry recruiters

agentic commerceAI paymentstokenisationanti-fraudOpenAIcard payments

More in Fintech

All Fintech news

OpenAI Launches ChatGPT Wallet for AI Agent Purchases With Card Tokenisation — HighRisk Media