Business

Mastercard CEO shares grave concerns over AI agentic commerce

Imagine a program you've never seen or used having access to your credit card and buying things on your behalf. That future may be arriving sooner than many people realize, and Mastercard's CEO says consumers should pay attention to the potential risks.

Michael Miebach, who leads one of the world's largest payment networks, told Yahoo Finance that agentic commerce will soon reshape daily life.

But his message focused less on opportunity and more on unresolved consumer protection challenges facing the payments industry.

Companies from Robinhood to Stripe are already building infrastructure for AI-powered credit card transactions. The technology is already here, but industry safeguards and consumer protections are still catching up.

Mastercard's CEO flags serious security gaps in AI-powered spending

Miebach described agentic commerce as the use case where artificial intelligence will "touch our lives the fastest and most broadly" during a recent appearance on Yahoo Finance's "Opening Bid" program.

But the Mastercard chief executive quickly pivoted to what worries him, posing a series of questions the industry has yet to answer.

"What happens if something goes wrong?" he asked, then he questioned whether AI agents would faithfully follow a consumer's spending instructions.

Miebach also raised doubts about identity verification, asking, "Is the agent actually what the agent claims to be?"

He described a phased trajectory from "assisted agentic payments first" to an eventual stage of "purely autonomous spending" with minimal human oversight.

Those concerns come as AI becomes increasingly embedded in everyday financial decisions. More than half of Americans used AI to help manage their finances over the past 12 months, according to Plaid's 2026 "State of Intelligent Finance" report.

How Robinhood and Stripe are building the AI payments ecosystem

Agentic payments are transactions that an AI agent initiates and completes on a consumer's behalf using secure digital card credentials and preset rules.

You could instruct an AI agent to track a product's price and automatically purchase it once the price drops below a certain threshold.

More AI:

The concept entered mainstream consumer finance on May 27, 2026, when Robinhood launched its agentic credit card for Gold Card holders.

The feature lets cardholders connect third-party AI agents to their Robinhood accounts and delegate purchasing decisions within preset spending caps and controls.

Stripe also expanded its Shared Payment Tokens in March to let AI agents initiate transactions using Mastercard- and Visa-issued agentic network tokens for security.

Visa has developed its own Trusted Agent Protocol, and both major card networks endorsed Google and Shopify's Universal Commerce Protocol at its launch.

The Consumer Bankers Association said in a January 2026 report that agentic payment tools carry "massive potential to disrupt the existing consumer payments landscape."

 Robinhood and Stripe push agentic payments forward as AI systems begin handling consumer transactions under preset spending rules and network security protocols.
Robinhood and Stripe push agentic payments forward as AI systems begin handling consumer transactions under preset spending rules and network security protocols.

Weiquan Lin/Getty Images

Rising fraud adds urgency to Mastercard's agentic payment warnings

The expansion into agentic payments is unfolding alongside a sharp rise in digital fraud, making Miebach's security concerns especially pressing.

Consumers lost more than $12.5 billion to fraud in 2024, according to Experian's 2026 Future of Fraud Forecast.

Nearly 60% of companies also reported an increase in their fraud losses between 2024 and 2025, the same Experian forecast indicated.

Institutional concern runs deeper still: 78% of financial institutions expect fraud linked to AI agents to rise significantly, Accenture research found.

One of the things about [large language models] that I find particularly concerning is that no matter what you ask it, it'll always come back with an answer that sounds authoritative, even if it's not.

Jodie Kelley, CEO of the Electronic Transactions Association, offered a more measured perspective in her January 2026 testimony before the House Financial Services Committee.

She noted that existing principles of authorization, consent, liability, and auditability still apply to agentic transactions, signaling the framework needs updating rather than replacement.

Mastercard's Agent Pay program targets the trust gap in AI transactions

Mastercard is not only identifying these risks, but also building infrastructure to address them, backed by a strong first-quarter financial performance.

Mastercard posted first-quarter 2026 net revenue of $8.4 billion, a 16% increase from the prior year, with adjusted earnings per share of $4.60, its April 30 earnings release confirmed.

Revenue from value-added services and solutions, Mastercard's highest-margin segment, climbed 22% during the quarter on a year-over-year basis at reported rates (18% on a currency-neutral basis).

The centerpiece of the company's agentic commerce response is Agent Pay, launched in 2025, which uses tokenization to replace sensitive card numbers with secure digital credentials.

Agent Pay also includes a "Know Your Agent" verification process that registers AI agents with unique identifiers, allowing banks to approve or block them.

Mastercard has partnered with Microsoft to integrate the framework into Copilot Checkout and is collaborating with OpenAI, Cloudflare, PayPal, and Stripe.

What autonomous AI spending means for consumer financial security

Agentic commerce raises a fundamental tension between convenience and oversight, one that Miebach argues is not new to the payments industry.

On "Opening Bid," he said consumers have always wanted simplicity, security, and a clear understanding of what happens when transactions go wrong.

Current safeguards like spending limits, manual-approval toggles, and tokenized credentials offer real protection, but they were designed for an assisted model.

The trajectory Miebach described toward fully autonomous AI spending will test whether those guardrails hold as the technology grows more sophisticated and widespread.

Related: AI CEO just made a wild prediction about AI agents

The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.

This story was originally published June 9, 2026 at 10:17 AM.

Get unlimited digital access
#ReadLocal

Try 1 month for $1

CLAIM OFFER