- Many employees secretly use AI tools despite company restrictions
- Employees sometimes use their personal devices to conceal AI use
- Larger organizations report higher levels of unauthorized AI usage
Artificial intelligence is becoming a routine part of office work, even when employees believe company policies prohibit its use.
New research from PagerDuty claims two-thirds (66%) of office professionals have used AI tools for work despite thinking those tools were not permitted.
The findings suggest unauthorized AI use is no longer isolated, particularly as workers become more confident in their own understanding of the technology.
Workers increasingly conceal AI use from employers
The study found that among employees who have used AI for work, many acknowledged taking actions that conflict with internal company rules.
Unauthorized use was more common at organizations employing at least 1,500 people, where 72% admitted using AI despite believing it was prohibited.
At smaller companies, the figure remained substantial at 60%, indicating that the practice extends across different workplace environments.
Secrecy often accompanies workplace AI adoption, as a third of AI users said they would deliberately avoid disclosing their use of AI to managers or supervisors.
About 30% cited restrictive company policies or concern about coworker reactions as reasons for keeping their AI use private, while 29% said uncertainty surrounding company rules contributed to their reluctance to disclose those activities.
Perceptions of inconsistent policy enforcement may also contribute to the use of AI in the workplace.
While 86% reported working at organizations with AI policies, 81% believed leadership operates under different standards.
That sentiment was especially common at larger organizations, where workers were more likely to feel executives received different treatment regarding AI-related decisions and policy compliance.
Most workers (72%) believe that they understand how to use AI for their jobs better than the teams managing AI governance.
At billion-dollar companies, that figure rose to 80%, while senior leaders were more likely than lower-level managers to express similar views.
With such confidence in their own AI judgment, workers will likely bypass formal restrictions, treating policy violations as reasonable workarounds.
Unauthorized AI activity often extends beyond simple tasks and 43% of respondents admitted to entering emails or work-related data into public AI systems.
Those tools operate outside internal corporate environments, creating potential concerns about how workplace information is handled after submission.
The sharing of sensitive information was not limited to routine communications, as more than a third of respondents said they had entered customer information into public AI systems.
Another 31% acknowledged uploading financial information, confidential company documents, or internal business strategies into those platforms.
Additional findings showed that 44% used AI tools to circumvent limitations in approved workplace software, while 38% shared AI-assisted work without disclosure.
In addition, some workers even access these AI tools or LLMs on their personal devices to conceal any trace.
More than half of those caught violating AI policies received informal guidance, while 48% faced formal disciplinary action.
This suggests that organizations are still struggling to balance enforcement with rapidly expanding workplace reliance on AI tools.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.