Surprisingly Useful AI Article Enhancements
Forty per cent of Australian office professionals have entered customer data into public AI tools such as ChatGPT, Claude or Gemini, while 28 per cent have input financial information or disclosed confidential company documents and strategies, according to research commissioned by PagerDuty.
The survey of 250 Australian office professionals at companies with a minimum annual revenue of $500 million, conducted by Wakefield Research, paints a picture of workplaces where AI use is widespread, often unauthorised and increasingly hidden from management.
Seventy per cent of Australian respondents had used AI tools or services at work despite believing they were not allowed to under company policies.
More than half (53 per cent) faced formal consequences such as a warning or disciplinary action as a result of that unauthorised use. A further 58 per cent received informal feedback or guidance, while 27 per cent experienced disapproval from coworkers.
Workers think they know better
A running theme through the findings is employee confidence. Three in four (75 per cent) Australian office professionals believe they understand how to use AI for their job better than the team responsible for managing AI at their company.
That sentiment held across all four regions surveyed, which also included Japan, the United Kingdom and the United States, with 1,250 people surveyed in total. The figure climbed to 80 per cent at billion-dollar enterprises.
Senior leaders (77 per cent) were more likely to feel this way than mid-level managers or those below (66 per cent).
A strong majority of Australian respondents (82 per cent) also believe their company’s leaders trust employees to work out which AI tools suit their day-to-day work for themselves.
Personal use comes first
The survey found AI adoption at work is being driven heavily by personal use. Among workers globally who have used AI in their work, 89 per cent first used an AI tool at work after trying it in their personal life.
In Australia, 58 per cent had done this multiple times, compared with 60 per cent globally. Regular AI users were much more likely (60 per cent) to have adopted AI tools at work after personal use than those who use AI only occasionally or less often (34 per cent).
Once workers start using AI on the job, it becomes their primary context for the technology. Those using AI for work spend more time with it on the job (76 per cent) than outside of work (24 per cent).
At home, 84 per cent of Australian respondents are using AI at least occasionally in their personal lives. The most common personal uses are research and learning (42 per cent), planning and organising schedules (34 per cent), writing or creating content (30 per cent) and hobbies and activities (30 per cent).
Why workers bring AI to work
Among those who adopted AI tools at work after personal use, 48 per cent in Australia did so because they became confident using the tool through personal use.
Others brought AI to work because of changing conditions: AI becoming more widely accepted in their workplace (39 per cent), their organisation formally approving or introducing AI tools (39 per cent) or a role change that made AI relevant to them (34 per cent).
Globally, 60 per cent are regularly using AI as part of their work responsibilities.
Related: Best Business Laptops for work & school
Related: Best Gaming Laptops
Related: Best Portable Laptop
Sensitive data going into public tools
The data security findings are among the most pointed in the survey.
More than a third (37 per cent) of Australian respondents have entered emails or other work-related correspondence into public AI tools that are not part of their company’s internal systems.
Beyond that, 40 per cent have entered customer data or information and 28 per cent have input financial information or disclosed confidential company documents and strategies.
Rules seen as inconsistent
One factor that may be fuelling unauthorised use is a perception that AI policies are applied unevenly.
While 82 per cent of Australian respondents work at companies where they believe AI policies exist, 83 per cent believe there are different rules or treatment of rules around AI for leadership compared with the rest of the company.
Hiding AI use
Over a third (36 per cent) of Australians who have used AI for their work would hide their AI use to avoid scrutiny from managers or leadership.
Another 30 per cent would hide it because company policies are restrictive or to avoid judgment from coworkers. Some 28 per cent are unsure whether their AI use is allowed. A quarter (25 per cent) claim they would not hide or avoid disclosing their AI use at all.
Separately, 45 per cent would rather use AI for work without telling anyone than risk being told they cannot use it.
Working around company tools
The survey found workers are using AI to work around existing company infrastructure.
Almost half (44 per cent) in Australia have used personal devices to complete work tasks or projects with AI. Thirty-nine per cent have turned to AI for work-related advice or help making decisions.
Forty-four per cent have used AI to get around limitations with existing company-approved tools. More than a third (34 per cent) have shared AI-assisted work without disclosing that AI helped.
AI restrictions pushing workers to look elsewhere
Seventy-two per cent of Australian respondents would be likely to look for a new job that offered better AI skills development.
More than two-thirds (78 per cent) believe their company’s restrictions or policies on AI usage are limiting their professional growth or career mobility.
That pressure could shift as more companies invest in training. More than half (53 per cent) believe their company is investing adequately in AI skills training and resources.
About the survey
The PagerDuty Shadow AI Survey was conducted by Wakefield Research among 1,250 office professionals working at companies with a minimum annual revenue of $500 million, excluding IT and technology roles, across Australia, Japan, the United Kingdom and the United States. Quotas were set for 500 respondents in the US and 250 in each remaining market. The survey was conducted using an email invitation and an online survey.
Last Updated on July 6, 2026 by Nick Ross
