PagerDuty's 2026 Shadow AI Survey — conducted by Wakefield Research among 1,250 office professionals at organisations with annual revenues of $500 million or more — has uncovered a striking disconnect between employee AI adoption and corporate governance. Two-thirds (66%) of office professionals have used AI tools at work despite believing they were not permitted under company policy, and many are feeding sensitive business data into public models in the process.

The survey covered non-IT and technology roles across Australia, Japan, the United Kingdom, and the United States. Its findings paint a picture of a workforce that believes deeply in the value of AI — and is increasingly willing to ignore the rules to use it. Left unaddressed, that gap generates measurable risks around data security, workforce trust, and talent retention.

Personal AI Adoption Is Driving Workplace Shadow AI

The path into the office often starts at home. Among workers who use AI for job responsibilities, 89% first adopted the technology outside of work before bringing it in. Outside of work, nearly all AI tool users (96%) say the technology has helped them in at least small ways, while nearly a quarter (24%) describe it as transformational — a figure that rises to 29% in the US.

Workers who use AI at home often become power users in the office: 67% of those who regularly use AI on the job have brought multiple personal AI tools into their professional workflow, compared to 47% of employees who use AI at work only occasionally. The biggest driver of workplace adoption was spotting clear opportunities to work more efficiently (45%), followed by growing workplace acceptance of AI (43%). Regardless of how adoption began, those using AI for work now use it more on the job (79%) than outside of it (21%).

---

Sensitive Data Is Flowing Into Public AI Models

The data employees are entering into public tools is where the real risk lies. 43% of respondents have entered work-related correspondence into public AI tools such as ChatGPT, Claude, or Gemini that are not part of their company's internal systems — with that figure rising to 51% in the UK and 50% in Japan. More than a third (34%) have entered customer data or information, and 31% have disclosed financial information or confidential company documents and strategies into public models.

The problem is more acute in smaller organisations. At companies with fewer than 1,500 employees, 40% of workers have entered customer data into public AI models, compared to just 27% at larger organisations. Meanwhile, the secrecy around this behaviour is notable: 39% of survey respondents said they would rather use AI without telling anyone to avoid being told it is not allowed — a figure that climbs to 47% at companies with over $1 billion in revenue.

---

Policy Inconsistency Is Fuelling the Problem

Employees aren't simply reckless — many feel the rules don't apply equally. While 86% of respondents believe their organisation has formal AI policies in place, more than four in five (81%) believe those policies are applied differently to leadership than to the rest of the workforce. This perception is even stronger among mid-level managers and below (85%) and in larger companies (85% by both revenue and headcount).

The consequences for those caught are significant: 53% of employees who used AI outside of company policy received informal feedback or guidance, while 48% faced formal consequences such as warnings or disciplinary action. Separately, 77% of respondents believe company restrictions on AI usage are limiting their professional development or career progression — a retention risk that organisations cannot afford to overlook. The survey's central message for leadership is that the goal should not be to slow down AI adoption, but to redirect employee energy toward governed, approved platforms that deliver both productivity and security at scale.