Report: AI Tools Expanding Corporate 'Shadow IT'
A new benchmark report from SaaS management platform Torii finds that the proliferation of AI tools is accelerating SaaS sprawl and expanding 'shadow IT' within companies. According to the report, 61% of applications are unmanaged, increasing governance and security risks as employees adopt new AI-powered SaaS products without official oversight.
- The term "shadow AI" describes the use of AI tools by employees without the formal approval and oversight of IT departments, a modern extension of the long-standing "shadow IT" problem. - Unsanctioned AI use poses greater risks than traditional shadow IT because data fed into public generative AI models can be permanently stored, used for training, and become irretrievable, creating indefinite data exposure. - High-profile instances of shadow AI risks include Samsung employees accidentally leaking proprietary source code and internal meeting notes to ChatGPT and Amazon discovering its internal data mirrored in ChatGPT responses. - Common examples of shadow AI in a corporate setting include developers using unapproved AI for code generation, support agents pasting customer data into public chatbots, and marketing teams using personal AI accounts to analyze internal data. - Beyond data leaks, risks include regulatory compliance violations with frameworks like GDPR and HIPAA, loss of intellectual property if trade secrets are input into models, and potential legal issues if AI-generated content is inaccurate or biased. - The drive for productivity fuels shadow AI, with one survey finding that 46% of employees would continue using AI tools even if their organization banned them. - To mitigate these risks, experts recommend establishing clear AI usage policies, providing sanctioned and secure AI alternatives, and utilizing SaaS Management Platforms (SMPs) to discover and monitor unapproved AI tool usage across the network. - The financial impact of AI-driven SaaS sprawl can be significant, with some enterprise customers finding that the adoption of AI features across their software stack can double or triple their aggregate SaaS costs.
