On September 9, Visa released the Visa Trust Index for agentic commerce, a new benchmark that tracks how comfortable consumers are with software agents that search, select, and pay without constant supervision. The first edition, based on a Harris Poll survey of 2,065 U.S. adults run in late May, puts hard numbers on the distance between using AI and letting it spend.
Highlights#
- Seventy-two percent of U.S. adults have tried an AI assistant, but only 23 percent trust generative AI itself to execute a payment. People will delegate discovery all day; the wallet is the last thing they hand over.
- The hesitation softens when a familiar payments brand stands behind the transaction. Sixty-one percent said they would trust Visa to handle agent-initiated transactions, the highest score among the payments, technology, and social media brands tested. Among 18-to-34-year-olds the figure rose to 68 percent, and among frequent AI users to 71 percent.
- Visa group president Oliver Jenkyn called trust "foundational," arguing agents will only gain spending autonomy once payment experiences feel as secure as e-commerce and mobile wallets did in earlier shifts.
- Visa says it will rerun the index as sentiment evolves, and positions Visa Intelligent Commerce, its bundle of standards, identity checks, authentication, and consumer controls, as the infrastructure meant to close the gap.
- Earlier Visa research covering the U.S., Australia, and New Zealand adds context: roughly two-thirds of consumers use or would use AI shopping agents, nearly nine in ten want transparency into how agents decide, and about half would stop using an agent if that control disappeared.
Why it matters#
The index names the real bottleneck in agentic commerce: settlement trust, not agent capability. Curiosity becomes revenue only when someone can prove who the agent is, what it was authorized to do, and where the money goes, which is exactly the evidence chain that escrow and reputation rails exist to provide.