Only 16% of US consumers trust AI to make payments on their behalf. That number alone explains why the "agentic commerce" future everyone keeps predicting remains stuck in pilot programs and demos.
The technical barriers are just as real. Developers trying to build AI agents that can browse, select, and purchase products autonomously are hitting walls at every layer of the payment stack. Card issuers won't talk to individual developers. Stripe requires 3D Secure verification for off-session payments (meaning a human has to approve the transaction anyway). E-commerce sites actively block browser automation. And Amazon's recent legal action against Perplexity for unauthorized scraping has made it clear that retailers will fight back against bots on their platforms.
The Infrastructure Gap
The disconnect is striking: AI agents are genuinely good at product research and comparison shopping. They can parse reviews, compare specs, find deals, and match preferences. But the moment they need to actually complete a purchase, the entire payment infrastructure assumes a human is on the other end.
This isn't accidental. Payment systems were designed around human authentication - passwords, biometrics, "click the traffic lights" CAPTCHAs, and two-factor codes. These exist specifically to prevent automated transactions, which is exactly what a shopping agent needs to do.
Some companies are working on bridges. Stripe's ChatGPT checkout integration lets AI prepare a cart but requires human confirmation before payment. Virtual card providers are experimenting with API-first issuance for agents. The emerging pattern is "AI prepares, human approves" rather than full autonomy.
The Trust Problem Runs Deeper
Even if the technical barriers were solved tomorrow, the trust gap would remain. While 52% of consumers say they're comfortable sharing data with AI shopping agents, 83% express concerns about privacy and data misuse. People want AI to help them shop. They don't want AI to have their credit card.
The compromise that's gaining traction is granular permission systems: "auto-buy anything under $50 that matches my regular purchases, but ask me about everything else." It's less dramatic than the fully autonomous shopping agent vision, but it's the version that might actually ship.
McKinsey projects agentic commerce could generate $3-5 trillion globally by 2030. But the EU AI Act takes effect in August 2026 with potential fines up to 7% of global revenue for non-compliance, and regulations around AI-completed transactions are still being written. The legal framework hasn't caught up to the technology, and until it does, the safest AI shopping agent is still one that asks you before it buys.