The AI Daily Five · No. 27 · Monday 16 November
Who Holds the Wallet When AI Pays?
From ChatGPT’s new wallet and ten million AI-agent payments on the XRP Ledger to Jensen Huang’s resignation, Bitcoin’s ETF outflows and Coinbase’s limits on AI trading, today’s edition asks who holds the wallet when AI pays.

1
OpenAI’s Wallet Lets AI Agents Make Purchases
ChatGPT’s new payment feature connects autonomous shopping to existing credit card networks. Every transaction still requires human approval, but the company is extending AI’s role from recommending products to buying them.
OpenAI has introduced ChatGPT Wallet, allowing its AI agents to complete purchases through existing payment networks without accessing customers’ actual card details.
Announced on 13 November, the service was developed with American Express, Mastercard and Visa. It is rolling out initially to US ChatGPT Pro subscribers aged 18 and over through OpenAI’s dots personal agents.
The process begins when an agent identifies a product or service the customer wants to purchase. Before payment, the customer reviews the merchant, items, total and selected card, then authorises the transaction using biometric verification such as Face ID or Touch ID.
The payment network then generates a substitute credential instead of exposing the customer’s real card number. The transaction uses the existing card account, preserving applicable rewards and protections for qualifying disputes.
OpenAI says merchants can use their existing checkout systems, although they may need to adjust fraud-detection procedures to recognise legitimate purchases made by AI agents.
The announcement arrives as blockchain developers experiment with alternative payment systems. Protocols such as x402 allow software agents to pay for individual digital services, frequently using cryptocurrency or stablecoins.
OpenAI’s approach follows a different route. Rather than requiring consumers to adopt unfamiliar payment instruments, it connects AI purchasing to the cards they already use.
But compatibility remains a practical obstacle. OpenAI acknowledges that some merchants may reject agent-generated transactions, while authorising a payment does not guarantee that an order has been accepted.
2
XRP Ledger’s AI Payments Pass 10 Million
Software agents are increasingly paying for computing services through blockchain networks. The transaction numbers are impressive, but their financial value tells a more complicated story.
AI payment infrastructure company t54 has reported more than 10 million transactions processed through its x402 facilitator on the XRP Ledger (XRPL), demonstrating growing use of blockchain-based micropayments between software agents.
The company confirmed the milestone on 2 October, reporting that activity had increased from approximately one million transactions to more than 10 million in less than three months.
In an interview highlighted by the XRPL Foundation on 15 November, t54 founder Chandler Fang discussed the rapid expansion of agent payments. The interviewer cited approximately 11.7 million transactions, according to reporting by U.Today.
The technology allows AI agents to purchase services such as computing resources, data access and model inference without requiring a person to approve each individual micropayment.
The x402 protocol integrates payment into ordinary web requests. When a service requires payment, an agent can respond with the appropriate transaction, using XRP or Ripple’s dollar-backed stablecoin RLUSD.
However, transaction counts require careful interpretation.
Earlier XRPL AI Hub figures reported in September recorded approximately 4.5 million payments but only 5,836.71 XRP and 4,125.29 RLUSD in cumulative settlement amounts.
Those figures illustrate how millions of transactions can represent relatively modest financial activity.
The development nevertheless provides evidence of an emerging market for machine-to-machine services. Developers can charge small amounts for individual tasks rather than require conventional subscriptions.
What remains less certain is whether the activity will translate into substantial commercial demand outside specialised developer ecosystems.
3
Jensen Huang Resigns. What Happens to Nvidia?
The co-founder who led Nvidia for 33 years has left following allegations concerning his personal relationships. His departure puts succession, governance and the company’s extraordinary influence over AI infrastructure under scrutiny.
Jensen Huang has resigned as Nvidia’s president, chief executive and board member, ending more than three decades leading the company that supplies much of the computing infrastructure behind modern artificial intelligence.
According to a filing with the US Securities and Exchange Commission, Nvidia’s board accepted Huang’s resignation on 11 November. The company publicly announced the transition the following day.
Chief financial officer Colette Kress has been appointed interim CEO while Nvidia searches for a permanent successor. She will continue serving as CFO during the transition.
The departure followed reports concerning alleged extramarital relationships, including allegations involving a woman described in US media as having links to Chinese intelligence.
The allegations have prompted questions about company governance and potential national-security implications. They should not, however, be treated as established findings of espionage or wrongdoing.
Nvidia’s board had previously announced an independent investigation, retaining external legal advisers to examine the circumstances.
The leadership change comes at a particularly sensitive moment for the semiconductor industry. Nvidia’s graphics processing units underpin major AI training and inference operations, making the company’s commercial decisions important far beyond its own shareholders.
Kress joined Nvidia in 2013 and has overseen its financial operations through much of its expansion into AI infrastructure.
Her appointment offers management continuity, but Nvidia now faces the challenge of replacing a founder whose strategic decisions helped define the company’s position in the global technology industry.
4
Bitcoin Slides as Investors Withdraw $1.9 Billion
Bitcoin has fallen towards $62,000 following sustained withdrawals from US exchange-traded funds. The decline contrasts with growing interest in blockchain payments and other cryptocurrency applications.
Bitcoin fell towards $62,000 over the weekend as selling pressure spread across digital-asset markets, adding to losses following several consecutive sessions of withdrawals from US spot Bitcoin exchange-traded funds.
The cryptocurrency traded near $62,081 on 15 November before recovering some of its losses, according to market reporting from PrimeXBT.
The decline followed a difficult period for cryptocurrency investors, with falling technology shares and concerns over US economic and political conditions weighing on sentiment.
Data from Farside Investors shows that US spot Bitcoin ETFs recorded approximately $1.91 billion in combined net outflows across seven consecutive trading sessions between 5 and 13 November.
These figures provide evidence of declining investor appetite through regulated investment products, although they do not establish a single cause for Bitcoin’s price movements.
The sell-off comes as financial institutions and technology companies continue developing new applications for digital payments.
AI agents can now access blockchain-based micropayment infrastructure, while conventional payment networks are entering agentic commerce through partnerships such as OpenAI’s recently announced ChatGPT Wallet.
Yet technological adoption and cryptocurrency prices do not necessarily move together.
Growing use of blockchain infrastructure may create demand for particular services without producing a corresponding increase in the market value of the underlying tokens.
The latest decline is therefore a reminder that claims about the expanding utility of digital assets must be assessed separately from their performance as speculative investments.
5
Coinbase Adds Limits to AI-Powered Trading
New safeguards can prevent autonomous trading systems from placing further orders after a portfolio loss. But stopping an AI agent does not automatically eliminate investors' existing market exposure.
Coinbase has introduced new controls for AI-powered trading, allowing customers to restrict the size and frequency of transactions carried out by software agents.
The cryptocurrency exchange announced Agent Controls on 11 November, extending its Coinbase for Agents service with safeguards designed to operate independently of the AI systems placing trades.
Customers can set a maximum amount for each transaction and establish a limit on total trading volume over 24 hours.
A separate auto-stop feature pauses new agent trades when a portfolio falls beyond a percentage-loss threshold selected by the user.
For example, a customer could configure the system to stop further AI-generated trading activity after a 5% portfolio decline.
Importantly, the controls operate on Coinbase’s servers rather than within the agent itself. This means the AI cannot independently remove or modify the restrictions.
However, a trading pause should not be confused with closing existing positions or cancelling outstanding orders.
As CryptoSlate reported on 14 November, existing orders can remain active after the system stops placing new trades.
That distinction matters during volatile trading periods, when positions can continue to gain or lose value regardless of whether an automated system has been instructed to stop.
Coinbase also plans to introduce more granular product permissions, allowing customers to choose which categories of assets their agents may trade.
The company warns that AI agents may make mistakes and that users remain responsible for the risks associated with transactions they authorise.
The announcement illustrates a practical challenge for financial automation: limiting an agent’s activity is considerably easier than guaranteeing a financial outcome.
The AI Daily Five