CIOs Push 5 Steps to Curb AI Sprawl as 88% of Organizations Use AI
Updated
Updated · Computerworld · May 18
CIOs Push 5 Steps to Curb AI Sprawl as 88% of Organizations Use AI
8 articles · Updated · Computerworld · May 18
Five measures are emerging for CIOs to contain AI sprawl: build real-time visibility, enforce guardrails, formalize employee-built tools, create internal AI infrastructure, and tighten vendor oversight.
88% of organizations already use AI in at least one business function, while unsanctioned usage often exceeds approved deployments by several multiples and spreads through assistants, no-code apps, scripts, agents, and embedded software features.
That fragmented adoption leaves IT with weak visibility and faster-rising risk, including sensitive data leaks, hallucinated outputs, unclear tool ownership, and AI costs that are hard to track to business value.
CIOs are shifting from blocking AI to enabling safer experimentation, using telemetry, identity data, registries, hosting environments, and AI-specific procurement and contract checks to keep innovation from moving further underground.
How can companies prevent autonomous AI agents from being hijacked for cyberattacks or causing catastrophic errors?
What hidden financial time bombs are ticking inside your company's unmanaged, employee-built AI tools?
In a zero-click AI world, how will businesses survive when search engines no longer send them traffic?
AI Agent Sprawl in 2026: Managing the Explosion, Governance Gaps, and Regulatory Risks
Overview
As of May 2026, enterprises are experiencing a rapid explosion in AI agent adoption, which is transforming business operations but also creating major challenges. The speed of this growth is outpacing companies’ ability to monitor and manage these agents, leading to a critical visibility gap and substantial business risks. This situation highlights the urgent need for robust governance and security frameworks. Organizations have shifted from early excitement to focusing on practical concerns like cost control, operational reliability, and especially governance, as the lack of mature oversight makes it difficult to ensure safe and effective use of AI agents at scale.