AI agents are a weekly habit, but trust stops at payment: what SMBs should do

AI agents have become an everyday habit: 68 percent of U.S. adults now use one at least weekly for a personal task. The same people stop short the moment an agent needs their card details or personal data. That gap is the central finding of a SEON survey released on November 2, and it matters to any business planning to roll out a chatbot, a voice agent or automated customer handling.

What the survey found

Fraud-prevention company SEON surveyed 2,026 U.S. adults between October 2 and 6, 2026. The key numbers:

• 68 percent use an AI agent weekly for personal tasks.
• 50 percent are uncomfortable letting an agent pay with their credit card details; the figure is 65 percent among Boomers and 34 percent among Millennials.
• Only 15 percent would share financial information with an agent, while 38 percent would share location, 34 percent purchase history and 34 percent calendar access. 32 percent would share nothing at all.
• 65 percent say the AI company behind the agent should be responsible for a fraudulent transaction; 14 percent would blame themselves, 11 percent the merchant and 10 percent their bank or card provider.
• 43 percent lack confidence in judging whether an agent’s request is legitimate, and 39 percent do not trust agents to recognise risky situations.
• The strongest trust builders: human support when something goes wrong (50 percent), the ability to reverse mistakes (42 percent) and strong identity verification (39 percent). 23 percent say nothing would increase their trust.

SEON CEO Tamas Kadar put it plainly: businesses need to tell a helpful agent from a harmful one before money moves.

Why this matters for SMBs

The sample is American, but the lesson travels. Customers are used to automated service and expect their suppliers to offer a fast digital channel too. At the same time, they hesitate exactly where a business has the most money and risk at stake: payment, identification and data handling. Deploying an AI agent without addressing both needs together means either losing the customer at the critical step or taking on fraud risk.

The liability question is instructive as well. Two thirds of respondents would blame the technology provider, but in practice the customer deals with the company that serves them. It is the SMB’s reputation that takes the hit, even when the failure happened at a vendor.

Practical steps

1. Separate the tasks. Let the agent handle information, bookings and order tracking; keep payment, contract signing and data changes as separate, confirmed steps.
2. Build in a human exit. Half of respondents rank it as the top trust factor: every conversation should be one step away from a person.
3. Make actions reversible. Cancellation, changes and refunds need a clear path, and the agent itself should explain it to the customer.
4. Verify identity on sensitive requests. On inbound calls and in chat alike, confirm who you are talking to before changing data or moving money.
5. Ask for minimal data. A third of customers would share nothing; the agent should request only what the task requires.
6. Log and monitor. Every agent interaction should be traceable, with suspicious patterns routed to human review.

An agent designed for trust

The numbers suggest the question is no longer whether customers use AI agents, but whether they trust yours. That is why BoostYourBiz’s AI Voice Agent and AI Chatbot are built this way from the start: they resolve routine questions and bookings on their own, verify identity at sensitive steps and hand over to a person in one step. If you would like to see how this would fit your processes, book a consultation.

Source: SEON / GlobeNewswire – “New SEON Survey Finds U.S. Consumers Are Embracing AI Agents, but 65% Would Blame the AI Company if One Commits Fraud”, November 2, 2026.