Agentic AI
,
Artificial Intelligence & Machine Learning
,
Governance & Risk Management
Executives Are More Likely to Use Unapproved AI Tools Than Their Teams
The term “shadow AI” conjures images of rogue junior employees feeding company data into their personal Claude and ChatGPT accounts when no one’s looking. But CIOs may be surprised to learn that the most frequent users of unauthorized artificial intelligence tools aren’t rank-and-file employees.
See Also: Edge Transformation: Top 5 SASE Predictions and Trends
They are executives in the C-suite, including CIOs.
According to a recent report from TrustedTech, 67% of senior decision-makers are using shadow AI tools, while just 31% of their employees admit to using unauthorized AI applications.
For C-level executives, that number jumps to nearly 73% admitting to using AI tools that the organization has not sanctioned, compared with 36% of entry-level staff. The survey, conducted in March through Censuswide, polled 2,001 U.K. and U.S. employees in junior and senior roles across organizations.
“The first thing CIOs need to accept is that this isn’t primarily a rank-and-file problem. It’s a leadership problem,” said Andy Nolan, vice president for technology at TrustedTech. “So, if a CIO’s governance strategy is built around monitoring junior staff, they’re pointing the telescope in the wrong direction. The exposure is concentrated at the top.”
The data shows it isn’t an awareness problem. It’s a trust problem.
When asked if they were concerned about shadow AI use in their organization, 56% of decisionmakers said yes, and 42% said they believed that their employers were monitoring their AI use. Only 31% of those they manage were concerned about shadow AI use, and 23% were worried about monitoring from above.
When asked why they used unapproved AI tools anyway, senior decision-makers said they weren’t bypassing approved tools because those tools are inadequate, but rather because they felt they were being surveilled. Only 28% said they found unapproved tools to be more efficient than the approved ones, and 29% said limitations in access were a problem.
But worries that the organization could see how often they were using the AI tools, and that it could have an impact on their career, was cited by 24% of decision-makers, 23% said they were worried AI monitoring would raise doubts about their abilities, and 23% said they were worried about being monitored by their manager.
CIOs should pay attention to the seeming contradiction here, Nolan said. While nearly two-thirds of senior decision-makers say they’re concerned about shadow AI use by employees, despite using it themselves, it’s not hypocrisy but a symptom of speed outpacing policy.
“Executives are under pressure to move fast, and when the approved tools don’t exist yet or don’t do what they need, they route around governance. That’s a completely human response, but it’s also exactly how you end up with sensitive data sitting in tools IT has no visibility into,” he said.
The leadership-heavy use of shadow AI also exposes another problem: The people most confident in AI are not necessarily the people most in need of support.
The findings reveal a stark contrast between the confidence level of senior and junior employees when it comes to using AI tools. Nearly 80% of senior decision-makers said they feel confident using AI, while 43% of their employees said the same. In the C-suite, 92% of executives feel confident. In entry-level roles, only 41% were confident in their AI capabilities.
That misalignment means that those who are setting policies and deploying tools may not be offering those who are using them the capabilties they need to generate real value. Only 23% of employees said that their AI skills were from employer training. Almost 55% of those below decision-maker level said they were self-taught or had learned from online resources such as blogs or YouTube.