Skip to content
IIAIT

AI Today · 14 November 2026

ChatGPT Wallet lets agents complete purchases, and the Fed's market contacts name AI as the top financial risk

OpenAI starts a limited US test of agent-made purchases with a confirmation step for every order, the Federal Reserve publishes its autumn financial stability assessment, and security researchers show how many apps built with an AI coding tool were left open to anyone.

  1. 01

    OpenAI launches ChatGPT Wallet so its dots agents can pay for purchases after user approval

    On November 13, 2026, OpenAI introduced ChatGPT Wallet, which lets its background agents, called dots, help choose products and complete checkout. Access is initially limited to a number of US ChatGPT Pro subscribers aged 18 or older with a US-issued Visa, Mastercard or American Express card. Before each purchase the user reviews the items, the total and the card, then approves with Face ID or a fingerprint. The card is charged through a payment token, so neither the agent nor the merchant sees the card details, and OpenAI says card rewards and dispute rights still apply.

    What it means for you

    For now every purchase still needs your explicit approval, which is the right default for work spending. If you plan to let agents buy on a company card, agree on limits and record-keeping first, and expect some retailers to block agent checkouts.

  2. 02

    Federal Reserve report: AI is the most-cited risk among its market contacts

    The Federal Reserve Board published its November 2026 Financial Stability Report on November 13. In a survey of 23 market contacts carried out by the New York Fed in September and October, AI was the most frequently cited risk to US financial stability, up from third place in the spring. Respondents pointed to three concerns: high valuations of AI-related stocks, cyberattacks made easier by AI or damage caused by frontier models themselves, and complex debt-based financing of the AI buildout. The Fed notes that the survey reflects respondents' views, not its own, and the report also includes a separate analysis of leverage in data center construction.

    What it means for you

    This is not a rule or a warning about any product, but it shows where financial professionals now see risk. For finance and accounting teams, it is a reasonable prompt to check exposure to AI-dependent vendors and to review how AI-assisted phishing and fraud are handled in your controls.

  3. 03

    Researchers find more than 35,000 Grok Build apps open to anyone, some holding internal company data

    Security firm Claroc reported that of about 37,500 apps it found online under the grok.me domain, 35,578 opened for anyone without signing in, and it estimates roughly one in ten held sensitive, internal or personal data. Engadget reports that affected organizations included six Fortune 500 companies, and that only one exposed app required a password. SpaceXSI says Grok apps stay private until their creator shares them and removed 24 apps built by its own staff, while the researchers say the paid account they tested had link sharing preselected. The report was published on November 10, shortly before this brief's usual window.

    What it means for you

    If people in your company build tools with AI app builders, find out which ones they use and check that apps holding internal data require sign-in. Open the shared link in a private browser window to see what an outsider would see, and unpublish apps that no one owns.