85% of IT teams claim every AI agent is under control, but only 42% actually know who owns them.

IT TEAMS CLAIM CONTROL OVER AI AGENTS: A CLOSER LOOK

Recent research reveals a striking confidence among IT teams regarding their management of AI agents. According to a survey conducted by Ivanti, a significant 85% of IT professionals assert that every AI agent within their organization has a named owner. This assertion reflects a strong belief in their ability to maintain control over AI technologies being utilized in the workplace. However, this confidence raises questions about the actual visibility and governance of these AI agents, particularly in light of the complexities introduced by rapid technological advancements and the growing prevalence of AI in various applications.

THE DISCONNECT BETWEEN AI AGENT OWNERSHIP AND CONTROL IN IT TEAMS

Despite the overwhelming majority of IT teams claiming control over AI agents, a concerning disconnect emerges when examining actual ownership clarity. The same Ivanti survey indicates that only 42% of IT professionals can definitively identify who owns these AI agents. This 43-point gap highlights a significant oversight in governance frameworks designed to manage AI technologies. It suggests that while IT teams may feel confident in their control, the reality is that many AI agents operate without clear ownership or accountability, potentially leading to risks and inefficiencies within organizations.

HOW ORGANIZATIONAL LEADERS ARE HIDING AI USAGE FROM IT TEAMS

The research also uncovers a troubling trend among organizational leaders who are increasingly concealing their AI usage from IT teams. The survey found that 42% of leaders admit to hiding their AI activities, a stark contrast to the 23% of all other employees who do the same. This secrecy is often justified by leaders seeking a "secret advantage," as noted by 52% of those who conceal their AI usage. Such behavior not only undermines the IT teams' ability to effectively manage AI agents but also creates an environment where risks associated with unmanaged AI can proliferate.

THE RISKS OF UNMANAGED AI AGENTS IN ENTERPRISE ENVIRONMENTS

The implications of unmanaged AI agents are significant, particularly in enterprise environments where data security and compliance are paramount. Sam Evans, CISO of Clearwater Analytics, expressed grave concerns regarding the potential misuse of customer data by employees who might inadvertently feed sensitive information into AI systems that are not properly managed. The risk of data breaches and the loss of intellectual property is heightened in scenarios where AI agents operate without oversight. As the landscape of AI continues to evolve, organizations must prioritize governance and accountability to mitigate these risks effectively.

AI AGENTS IN THE WILD: THE CHALLENGE OF SHADOW AI DISCOVERY

The challenge of identifying and managing AI agents extends beyond ownership and control; it encompasses the phenomenon of 'shadow AI.' As highlighted by industry experts, AI is now embedded in numerous applications and tools that employees interact with daily. For instance, Prompt Security's CEO, Itamar Golan, reported that their organization sees approximately 50 new AI applications emerge daily, with over 12,000 already cataloged. This proliferation of AI technologies complicates the discovery process, as many organizations struggle to keep track of the AI agents in use. The situation is exacerbated by the fact that around 40% of these applications default to training on any data provided, raising concerns about the potential exposure of sensitive information. The need for robust strategies to discover and govern these shadow AI agents is more critical than ever.