A recent study conducted by Cybersecurity Insiders in 2026 revealed that many major corporations are facing challenges in overseeing and managing artificial intelligence (AI) systems operating on their networks. The research, which surveyed 235 Chief Information Security Officers (CISOs), Chief Information Officers (CIOs), and other senior security executives in the US and the UK, indicated that a significant number of AI tools are being implemented without proper authorization.
The report uncovered that 75% of organizations have encountered unauthorized “Shadow AI” tools functioning within their networks, some of which have access to sensitive information. Despite this, only 16% of companies effectively regulate this access, even though 71% of CISOs acknowledge that AI systems have entry to crucial business operations.
One of the major concerns highlighted in the survey is the lack of visibility regarding AI identities. A staggering 92% of organizations admitted to not having complete oversight of their AI systems, and 95% expressed doubts about their ability to detect malicious activities carried out by AI agents. Merely 5% of respondents stated they were confident in their capability to contain a compromised AI system.
Security leaders pointed out that the swift and widespread adoption of AI tools, such as copilots, posed a significant challenge due to their autonomous nature, which makes them challenging to monitor using conventional security measures designed for human users. The report emphasized that 86% of leaders do not enforce access protocols for AI, and only 25% utilize monitoring mechanisms tailored specifically for AI.