Two-thirds of professionals use Shadow AI

According to a Forvis Mazars–ESSCA barometer published on October 1, 2026, two out of three professionals report unregulated uses of artificial intelligence in their company, a phenomenon referred to as "Shadow AI." This term describes employees using AI tools like ChatGPT without going through company validation channels, often to bypass what they perceive as unnecessarily complex work processes.

Risks of unregulated AI use

The practice has raised concerns among information technology departments, who highlight risks such as exposure of confidential data, circulation of unverified results, and entrusting sensitive information to external suppliers. These risks underscore the need for more structured oversight of AI tools within organizations.

Shadow AI beyond customer service

The phenomenon is not limited to customer service departments, where employees may spend significant time retrieving request histories, consulting procedures, and verifying information across multiple systems. In such scenarios, employees might opt to use AI tools to save time, even if these tools have not been officially adopted by the company.

Companies plan to invest in AI

The Forvis Mazars–ESSCA barometer also notes that 58% of respondents indicate their companies plan to increase investments in AI. However, the report suggests that before adding new technologies, companies should consider examining what employees are already trying to correct through unauthorized AI use.

Confusing speed with performance

The report highlights a potential confusion between execution speed and performance improvement. For instance, reducing the time spent preparing monthly reports using AI may appear beneficial, but if the reports are rarely used, the real progress would be in questioning their usefulness rather than focusing on speed.

Need for deeper organizational discussions

The report suggests that companies should engage operational managers and IT teams in discussions that go beyond budgetary arbitration or technical validation. These discussions should aim to understand the underlying issues that prompt employees to use unauthorized AI tools. This approach could reveal opportunities to streamline processes, review data circulation, and identify tasks that could be eliminated rather than automated.

Shadow AI as a reflection of organizational issues

While the report acknowledges the risks associated with Shadow AI, it also suggests that the scale of its occurrence should prompt general management to consider what motivates these practices. It argues that companies should not only focus on regaining control over AI tools but also reflect on the real functioning of their organizations and the burdens that hinder productivity.