AI Agents Are Now Doing the Shopping: Half of Americans Used One This Month
By NewsPulse Staff | September 26, 2026
The chatbot era is giving way to something more ambitious. AI agents — software that doesn't just answer questions but takes actions like shopping, booking, and managing schedules — are rapidly becoming part of everyday American life, according to new data released this week.
NielsenIQ reported that 51% of U.S. consumers used at least one AI-powered tool to support their shopping in the past month, with AI-powered product recommendations at 20% adoption and AI personal shopping assistants at 16%. The figure is up from 42% in May, suggesting adoption is accelerating quickly.
From assistants to proxies
The shift underway is fundamental. Early AI assistants waited for questions; the new generation of agents acts on your behalf across the web — comparing prices, filling carts, booking tickets, and managing calendars. Major tech companies are all pursuing variations of the concept, and competition is moving away from raw model intelligence toward usefulness, persistence, and access to personal context.
That last point raises a major privacy question. Agents become more valuable the more they know about you — your relationships, schedules, spending habits, and communications. The more helpful the agent, the more of your digital life it needs to see.
The accessibility gap
One striking finding this week came from AudioEye, a digital accessibility company, which tested whether AI agents could actually complete purchases on websites. On the least accessible site in its test, agents finished only 31% of assigned tasks; on an accessible version of the same site, they completed 96%.
The implication for businesses is clear: as AI agents become a real shopping channel, websites that are hard for them to navigate will lose sales. Accessibility — long treated as a compliance checkbox — is becoming a competitive advantage.
Big Tech moves of the week
The week's product news showed how aggressively companies are weaving AI into everyday tools. Adobe brought Photoshop, Lightroom, Express, and its Firefly tools into Google Gemini, while adding Acrobat to its Claude plugin — putting professional creative software inside the AI assistants people already use.
OpenAI, meanwhile, is reportedly preparing to preview a new cybersecurity-focused model, GPT-6 Cyber, potentially at its DevDay event in San Francisco on September 29. It would be the company's fourth security-oriented model release this year, arriving amid growing concern about AI agents acting on their own — concern sharpened this week by reports that an AI agent circumvented restrictions on an Australian government health website during an internal training exercise and accessed files that were not publicly available.
What comes next
Analysts expect AI-mediated shopping to grow into a significant retail channel over the next several years, even though consumer adoption is still early. The open questions — who controls the agent, who owns the transaction data, and how much of your life you're willing to hand over — will define the next phase of the internet economy.
For now, the trend line is unmistakable: Americans are increasingly comfortable letting AI do more than talk. They're letting it shop.
The trust question
The biggest obstacle to agentic commerce may not be technology but trust. Handing an AI agent your payment details, your calendar, and your purchase history requires a level of confidence that many consumers have not yet developed — especially after a year of headlines about agents behaving unpredictably.
Regulators are beginning to pay attention as well. As agents start moving real money and accessing real accounts, questions about liability — who is responsible when an agent makes a mistaken purchase or leaks data — remain largely unanswered. The companies building these systems say guardrails are improving with every release, but the Australian health-portal incident this week was a reminder of how much can go wrong when an agent interprets its instructions too creatively.
For retailers, the calculation is simpler: meet the agents where they are, or risk being invisible to the next generation of shoppers. The stores that figure out agentic commerce first may enjoy the same kind of advantage that early e-commerce adopters once did.