A gap is opening between how much AI enterprise workers use and how much they trust it to act on their behalf. New global research from TeamViewer puts hard numbers on a tension IT leaders have been circling for over a year.
The study, commissioned from Sapio Research across 4,200 respondents in nine markets including the UK, US, Germany, and Japan, found that 75% of employees now use AI at least once a day. Yet 61% still prefer AI to take no independent action, and 95% express some degree of concern about systems operating without a human in the loop.
The gap is sharpest where AI's value proposition is loudest. IT decision-makers are among the most willing to delegate — 99% said they would hand at least one task to AI, most commonly system updates (61%), common device fixes (53%), and performance adjustments (49%). But even within that group, the appetite for unattended autonomy is limited. Seventy-one per cent of those willing to allow more autonomous AI action still want it confined to clearly defined tasks.
The verification burden created by partial trust is already measurable. More than half of respondents (56%) say they often or always check AI outputs before acting on them, spending an average of two hours per week on that checking work.
What respondents describe wanting is not less AI but more legible AI: 40% want systems that detect and fix issues while keeping humans informed; 70% are comfortable with autonomous action as long as they can intervene. Guardrails matter more than raw capability — security and privacy protections (39%), notifications before major changes (36%), and visible activity logs (33%) ranked ahead of accuracy or speed as conditions for accepting greater autonomy.
Looking further out, the survey's picture of the autonomous workplace is steady-state rather than transformative. By 2030, 33% of respondents expect to still be doing most tasks with AI assistance, 28% expect to delegate more and shift to higher-value work, and only 2% expect their role to be eliminated.
TeamViewer CEO Oliver Steil argued the data points to a strategic constraint: "Organisations will not succeed by giving AI unlimited authority. They will succeed by building systems that operate safely, transparently and accountably within clearly defined guardrails."
The study, titled 'Path to the Autonomous Workplace', was conducted in April and May 2026. The full report is available at teamviewer.com.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
