AI Expands Shadow IT, Not Consolidation
Contrary to expectations that AI would consolidate SaaS stacks, it is accelerating SaaS sprawl and expanding shadow IT. A new report from Torii finds that 61% of applications are now unmanaged, increasing governance and security risks for enterprises as employees adopt new AI tools outside of official procurement channels.
- Research shows that 59% of employees use AI tools not approved by their employer, and 75% of them admit to sharing sensitive company data with these applications. This practice, known as "Shadow AI," creates significant security blind spots outside of corporate oversight. - The uncontrolled growth of applications, or SaaS sprawl, is accelerating, with organizations adding an average of nine new apps every month. This leads to significant budget waste, as roughly half of all SaaS licenses go unused and companies often pay for multiple redundant applications, with an average of 10 different project management tools per organization. - The data being shared with unapproved AI tools often includes sensitive information such as customer data (32%), internal documents (27%), financial information (21%), and proprietary code (20%). This exposure can lead to compliance violations with regulations like GDPR and HIPAA, as unvetted tools rarely have the necessary data processing agreements in place. - Senior leadership is a primary driver of this trend, with one study finding 93% of executives and senior managers use unapproved AI tools—the highest rate across all job levels. This creates a paradox where those responsible for compliance are often the most frequent users of unsanctioned software. - Marketing departments are significant contributors to shadow IT, with unapproved tools accounting for 35% of the average marketing tech stack. Teams often adopt their own AI-powered solutions for tasks like content generation, personalization, and data analysis to move faster than internal IT processes allow. - The use of shadow AI is often a response to inadequate official tools; while over half of employers provide approved AI software, only one-third of employees find these tools fully meet their needs to perform their jobs efficiently. This gap between the tools provided and the features employees require pushes them to find their own solutions.
