A nationally representative survey of 2,000 UK adults, commissioned by cybersecurity PR agency Eskenzi PR and conducted by Censuswide, found that 84% would be comfortable giving an AI agent moderate, low or no control over their day-to-day lives. More than a third (34%) would not give one any control at all.
AI agents can carry out tasks on a user’s behalf, potentially including booking travel, managing diaries, making purchases and accessing connected accounts. However, the findings suggest public acceptance depends heavily on oversight and clear limits on what these systems can do.
Just over a quarter of respondents (27%) would allow an AI agent to prepare tasks but would want to approve every action. A further 23% would let it handle routine tasks independently, while requiring permission for more important decisions. Only 11% would be comfortable granting high or very high levels of control.
Data security tops concerns
The most common concern about giving an AI agent access to accounts and permission to act was personal data being hacked or stolen, cited by 36% of respondents. This was followed by the agent making a mistake (31%) and having access to too much personal information (30%).
More than a quarter (28%) worried that an agent could spend their money or buy something they had not intended, while 22% feared it could communicate with someone without their knowledge.
Meanwhile, 14% were concerned that an AI agent could implicate them in something illegal. Among 18 to 24-year-olds, this rose to 21%.
Jonathan Armstrong, Partner at Punter Southall, said the findings highlighted the need for a debate about liability.
“Too many people assume that AI is the Wild West – legally that just isn’t the case. In most cases you’ll be responsible for the AI you use and the consequences – and sometimes even if it took you by surprise,” he said. “There are many great use cases for AI but they should all start with knowledge of the risks and doing what you can to reduce that risk.”
Public views on responsibility differed from that warning. Asked who should bear primary responsibility if an AI agent was hacked, tricked or did something it should not, 40% pointed to the company that created it. A quarter (25%) chose the company or service the agent was accessing, while 18% selected the individual using it.
Businesses face demands for oversight
The research also examined how consumers would respond if a bank, insurer or utility provider used an AI agent to advise its board on important decisions.
More than a third (37%) would want reassurance that a human always had the final say. Almost a third (31%) would want to know exactly which decisions involved the AI, while 24% would investigate how their personal data was protected.
One in five (20%) would consider switching provider, and 18% would be less likely to buy additional products or services. By comparison, 10% would view the use of boardroom AI as a positive sign that the company was modern and embracing technology. Just 6% said it would make them more likely to use the company.
Yvonne Eskenzi, co-founder of Eskenzi PR, said: “AI agents promise to take a lot of boring jobs off our hands, and people clearly see the appeal of that. But there’s a huge difference between asking AI to organise your diary and letting it spend your money, access your accounts or make important decisions for you.”
She added: “They want to know what it’s doing, they want someone to take responsibility when things go wrong and, when it really matters, they still want a human in charge.”
Age shapes willingness to hand over control
Almost half of respondents aged over 55 (47%) would give an AI agent no control over their day-to-day lives, compared with 21% of 18 to 24-year-olds and 18% of those aged 25 to 34.
Among 25 to 34-year-olds, 20% would accept high or very high levels of control, compared with 5% of over-55s.
When asked about predictions that advanced AI could threaten humanity’s survival by 2030, the most common response was greater concern about what humans would do with AI than about AI itself, selected by 22%. Only 9% said they were not worried about AI at all.
The findings point to a challenge for businesses introducing autonomous systems: demonstrating useful capabilities while giving users clear information about permissions, data protection and accountability.
The survey was conducted between 16 and 18 September 2026. The full report is available here.
