Most Americans would blame the AI company if its agent commits fraud

Most Americans who have an opinion say the AI company should take the blame if its agent commits fraud. A survey from fraud prevention firm SEON puts that figure at 65%. An AI agent is software that takes actions for you, such as completing a task or a purchase.

What the survey found

SEON polled 2,026 U.S. adults online from October 2 to 6, 2026. People are using agents for everyday tasks. They are much less willing to hand over money or sensitive data. Half say human support would make them trust AI agents more.

The blame question went to the 885 people who gave an opinion, not those who said "not sure." The margin of error is about 3 percentage points for questions asked of everyone.

Why businesses should care

SEON says banks, fintechs and payment providers will need to tell whether an agent is acting for a real customer or helping a fraudster. Spotting an agent is only the first step. The company says firms should also check identity, device and behavior clues. SEON sells fraud tools, so it has a stake in this message.

What is not known

The release does not say who else people would blame, or give exact numbers for most other findings. It also does not show the full questions.