A study by Cyberdom reveals that nearly one-third of employees have shared login credentials, keys, or access to work folders with AI models over the past six months. The research involved 1,310 employees and 490 managers and specialists in IT, HR, and information security.
Employees frequently upload sensitive materials to AI tools to speed up tasks, analyze large documents, or edit text. Reported uploads include commercial proposals (25%), financial reports (22%), work correspondence (21%), contracts (16%), and client databases (12%).
Corporate policies regarding AI remain rare, with only 5% of employees reporting the existence of written guidelines. Furthermore, 22% of staff stated that their companies prohibit the use of external AI services for any work-related materials, while 17% received only verbal instructions.
Security and IT professionals confirmed the risks, with 32% reporting that employees had shared passwords or access keys, and 28% noting the transfer of source code. Currently, 23% of companies do not monitor AI usage, and 16% of managers lack visibility into what data is being shared.
In response to data leaks, 23% of companies restrict access to AI services, while 20% review each incident individually. Only 8% of companies apply disciplinary measures, and 2% consider termination. About 30% of surveyed experts identified the need for internal corporate AI services as the top priority.
