Security Doubts Emerge as the Biggest Barrier to Workplace AI
Middle managers are nearly twice as likely as executives to want a person involved when AI resolves IT problems that touch security, data, or business operations, according to global research from TeamViewer. The split is 46 per cent against 24 per cent. Security, privacy, and risk concerns are the most cited barrier to AI adoption among IT decision makers, named by almost half of them.
Workers grow warier as systems take on more autonomy, and nearly a third say security risks or unauthorised access rank among their biggest worries about AI acting without a person in the loop. That figure reaches 33 per cent among middle and senior managers, and 23 per cent among executive leaders.
But the appetite for control varies sharply with seniority. Limiting what AI can reach would make 42 per cent of senior managers more comfortable with software fixing problems on its own, compared with 37 per cent of middle managers and 32 per cent of executives. Industry matters too, since almost half of workers in energy, utilities, and oil and gas want human involvement, against 28 per cent in marketing, media, and communications.
"Security cannot be added after autonomy is introduced," said Jan Bee, CISO at TeamViewer. "The more authority an AI system has, the more important it is that its actions are transparent and that clear boundaries define when human approval is required." He added that business leaders should not assume their own comfort with autonomy is shared across the organisation.
Strong privacy and security protections would make almost four in ten workers more comfortable letting AI fix problems autonomously. That figure rises to 46 per cent among senior managers and 43 per cent among middle managers, against 34 per cent of executive leaders. Within the region, 46 per cent of workers in Singapore agreed, compared with 42 per cent in Australia, 41 per cent across Asia Pacific, and 34 per cent in Japan.
And workers name the specific guardrails they want. More than a third want limits on what AI can access before it acts independently, while 73 per cent say they trust AI more when it shows what it is doing and why. Another 70 per cent are comfortable with AI taking action as long as they can step in when needed.
Doubt stays the default for many. Just over half of workers say they do not always know when to trust AI and when to check its output. Singapore sits at 50 per cent, level with the 50 per cent average across Asia Pacific, while Australia reaches 55 per cent and Japan 44 per cent.
Accountability is no clearer. Only 49 per cent say responsibility is clear when autonomous systems make mistakes, and the answers are spread thin. A quarter point to IT, 25 per cent take it on themselves, and 17 per cent believe nobody should carry it because the outcome came from a system.
So the pressure sits on frameworks rather than slogans. The study, run by Sapio Research, surveyed 4,200 managers and employees across nine markets, including Singapore, Japan, Australia, the United States, and Brazil. Fieldwork took place earlier this year, with an even split of managers and employees across industries and company sizes.
Middle managers are almost twice as likely as executives to want a human involved when AI handles sensitive IT issues
Security, privacy, and risk concerns are the top barrier to AI adoption for almost half of IT decision makers
73 per cent trust AI more when it explains what it is doing and why
Only 49 per cent say accountability is clear when autonomous systems get things wrong
Source: TeamViewer
