In a recent nationally representative survey comprising 2,000 adults from the UK, conducted by Censuswide and commissioned by the cybersecurity PR agency Eskenzi PR, intriguing insights regarding public perception of artificial intelligence (AI) have emerged. A significant 84% of participants expressed that they would be comfortable granting an AI agent varying levels of control over their daily lives, whether moderate, low, or none at all. Interestingly, over a third of the respondents—34%—indicated that they would not permit any control by AI systems.
The potential capabilities of AI agents are considerable; they could manage everyday tasks such as booking travel, overseeing calendars, making purchases, and accessing various connected accounts. However, the survey highlights that the extent of public acceptance is significantly contingent upon proper oversight and well-defined boundaries concerning the functionalities these AI systems can possess.
Respondents’ comfort levels with AI autonomy demonstrate a spectrum of preferences. Slightly more than a quarter (27%) of those surveyed would permit an AI agent to prepare tasks but would require personal approval for each action taken. Meanwhile, another 23% would allow the AI to independently complete routine tasks while reserving permission for more critical decisions. A mere 11% of participants were open to granting high or very high levels of control to AI systems, underscoring a cautious approach towards autonomy.
### Data Security Tops Concerns
When it comes to concerns surrounding the delegation of control to AI agents, issues of data security reign supreme. A considerable 36% of respondents revealed that their primary worry was the potential for personal data to be hacked or stolen. Furthermore, the possibility of the AI agent making a mistake was cited by 31% of participants, and 30% were anxious about the software having access to excessive personal information.
Additional concerns included financial implications, with 28% of respondents fearing that an AI could spend their money or make unwanted purchases. Moreover, 22% expressed anxiety over the agent potentially communicating with individuals unknown to the user. The apprehension escalates among younger demographics; 14% of all respondents were worried that an AI agent could implicate them in illegal activities, with this percentage rising to 21% among the 18 to 24 age group.
Jonathan Armstrong, a partner at Punter Southall, weighed in on the findings, emphasizing the necessity for discussions regarding liability in those contexts. “Many individuals mistakenly perceive AI as being above legal scrutiny; in reality, users often bear responsibility for the AI systems they engage with, even in unforeseen circumstances,” he said. Armstrong pointed out that while numerous beneficial applications of AI exist, they must be approached with a clear understanding of associated risks and diligent measures to mitigate them.
The public’s perception of accountability diverges somewhat from Armstrong’s cautionary stance. When respondents were asked who should hold the primary responsibility in cases where an AI agent is compromised, misled, or acts inappropriately, 40% attributed responsibility to the company that developed the AI. A quarter (25%) of respondents believed the company or service that the AI accessed should be accountable, while 18% felt the individual user should take on this burden.
### Businesses Face Demands for Oversight
The survey explored consumer reactions to the integration of AI agents in corporate advisory roles, particularly within banks, insurers, and utility providers. Over a third (37%) indicated a preference for assurance that a human would always have the ultimate decision-making authority. Approximately 31% wanted clarity about which decisions were influenced by AI involvement, and 24% would scrutinize how their personal data was safeguarded.
Consumer skepticism extends to the potential ramifications for businesses. One in five respondents (20%) voiced intentions to reconsider their relationship with service providers employing AI, with 18% indicating a reduced likelihood of purchasing additional services from such companies. Conversely, only 10% viewed boardroom AI utilization as a favorable indicator of a modern, technology-embracing business. Just 6% of respondents stated it would make them more inclined to utilize the company’s services.
Yvonne Eskenzi, co-founder of Eskenzi PR, remarked on this dichotomy, noting, “AI agents promise to alleviate mundane tasks, making them quite appealing to the public. However, a significant distinction exists between requesting AI to manage your schedule and allowing it to handle finances or make vital decisions.”
The survey results further reveal that willingness to delegate control to AI agents varies markedly with age. Nearly half (47%) of individuals aged over 55 expressed a reluctance to relinquish any control, in contrast to merely 21% of respondents aged 18 to 24 and 18% of those aged 25 to 34. Furthermore, among the 25- to 34-year-old demographic, 20% were open to high or very high levels of control, whereas only 5% of the over-55 group felt similarly.
As discussions unfold regarding predictions that advanced AI might jeopardize humanity’s survival by 2030, a predominant concern, selected by 22% of participants, revolved around the potential misuse of AI by humans rather than the technology itself. Only 9% of respondents reported no worries regarding AI.
The findings from this survey elucidate a significant challenge for organizations looking to implement autonomous systems: they must strike a balance between demonstrating the technology’s utility and ensuring that users receive transparent information concerning permissions, data protection, and accountability. The survey was conducted between September 16 and 18, 2026, and the comprehensive report is available for further review.