Technology

AI agents can shop for you. Is it safe?

AI “agents” promise to do everything from tidying up your email like Marie Kondo declutters a closet, to buying you a pair of heels based on your budget and style preferences.
The pitch sounds kind of effortless—until you remember that “effortless” is usually another word for “who’s actually pressing the button?”

Agentic commerce moves fast

Misryoum newsroom reported that technology experts are warning consumers to be careful. The concern is pretty direct: outsourcing key decisions to AI exposes people to communications errors, can end up costing money, and—maybe most worrying—could hand hackers the keys to consumers’ data.

This is especially true for something called agentic commerce, or relying on AI agents to make purchases for you.
Misryoum editorial team stated that it isn’t mainstream yet and “it’s pretty risky right now,” mainly because there still aren’t enough guardrails for people to feel comfortable with agents buying things on their own.
In one cautionary example, Misryoum newsroom reported the idea that an agent could theoretically go buy a car—except you probably shouldn’t hand over the credit card and hope for the best.

Even with that warning, big brands are pushing ahead.
American Express this week announced new services and protections for cardholders who make purchases using specified AI agents.
Misryoum analysis indicates that the setup includes verifying the identity of an agent when it makes a purchase, and the service “will protect eligible customers from charges related to AI agent error,” Amex said in a statement.

Where things can break

On the shopping side, Amazon’s agentic AI assistant, dubbed “Rufus,” can track the price of products on the online retailer’s platform, alert customers when the price hits a prescribed level and complete the purchase.
Walmart, meanwhile, has deployed what it calls a “conversational” AI agent named Sparky.
The company says it can help consumers find products, provide customer reviews and help with ordering.

Misryoum editorial desk noted that some people are already experimenting.
Roughly a quarter of Americans between the ages of 18 and 39 say they have tried using AI to research products or to shop, according to November data from market research firm Statista.
Still, “trying it” and “letting it spend your money unattended” are two different moods.

What could go wrong is not just theoretical.
Misryoum newsroom reported the story of Sebastian Heyneman, the founder of a San Francisco-based tech startup.
He instructed an AI agent to secure him a speaking opportunity at the World Economic Forum in Davos, Switzerland.
The bot succeeded—landing him a slot for $30,000, a fee he couldn’t afford.
The detail that sticks is how ordinary the setup sounded: he asked, the system delivered, and the bill arrived like a punchline.

Misryoum newsroom reported that Heyneman used a bot by Tasklet, a company that lets businesses automate routine business tasks with AI agents.
Andrew Lee, the founder of Tasklet, told Misryoum that problems can arise when a user prompt gives the AI conflicting instructions.
Lee also said agentic AI today is fully capable of shopping for people and doing “normal things consumers can do.” But just because tech can do something doesn’t mean it should be used that way, he warned.
“The specific use case of shopping is not a good thing to use these systems for — yet,” Misryoum newsroom reported Lee saying.
He added that the agents are “fundamentally hard to trust,” and he isn’t super comfortable with it personally—he likes to control where his money goes himself, and as a business, they don’t recommend it.

The threat model gets darker fast when credit cards enter the scene.
Misryoum newsroom reported Bretton Auerbach, founder of a New York-based tech startup, warning that bad actors can lure AI agents into turning over a consumer’s personal information.
In one example, he said if you give an agent your credit card and say, “Go to this website and buy me something online,” there are ways to trick the agent.
He described a scenario where it might mistake a legitimate website for a phishing website that says in big, bold, text, “Paste your credit card number here.”

To be fair, shoppers do like convenience.
And sometimes it works—prices drop, reminders pop up, purchases go through.
But if you’re sitting at your desk and you can hear the faint buzz of your laptop fan while you stare at a checkout page, you probably want to feel in control.
With agentic commerce, that feeling is… negotiable.
And as Misryoum editorial team keeps pointing out, the systems still aren’t quite built with the kind of guardrails that make “autonomous” feel automatically safe.

Valve’s Proton 11 beta boosts Windows games on Android

Apple Week Deals: $150 Off M5 MacBook Air, More

Geely’s Galaxy Starshine 7: $16,550 EV that feels premium

Back to top button