---
title: AI governance becomes unmanageable burden
url: https://www.elseif.net/ai-governance-becomes-unmanageable-burden
published: 2026-09-30T15:09:06+00:00
language: en
section: Models
source: https://computersweden.se/article/4227410/ai-styrning-haller-snabbt-pa-att-bli-en-ohanterlig-borda.html
organizations: One Trust, Boston Consulting Group, Flexera, Infosys, Tata Consultancy Services, Computer Sweden
publisher: elseif
---

# AI governance becomes unmanageable burden

AI governance is rapidly turning into an unmanageable burden for corporate leaders. Recent surveys show that four out of five senior decision makers, including CIOs, CISOs and CDOs, now spend a larger share of their workday handling AI risks. On average their working time has grown by 26 percent because of this extra responsibility. One Trust’s platform for AI governance published a new survey confirming that AI usage in firms has increased sharply and that risk‑aware action plans are replacing the old “anything goes” approach. The same study found that 86 percent of organisations have reported AI‑related incidents and that more than a quarter experienced two or more incidents where AI systems took unauthorized actions. Experts say the surge is driven by the sheer volume of AI being introduced, especially by non‑IT staff who are now allowed to build their own applications.

This citizen‑development trend forces governance to keep pace, creating an exponential growth in the number of uncontrolled AI tools. Security, compliance and risk management have historically only needed to steer people, but AI moves at machine speed and can create problems far faster than humans can react. Technical architects at Tata Consultancy Services note that much of the additional time spent on AI governance is devoted to tracking shadow‑AI rather than monitoring approved tools. When employees find a tool that works for them they do not wait for formal approval; governance must scramble to put controls around untested solutions. Simply reacting to incidents is not the same as building effective controls; time spent investigating unauthorized tools is fire‑fighting, not true governance.

Organisations that take AI governance seriously invest in architecture, data provenance, model versioning, escalation paths and audit trails before deployment. Awareness alone is not enough; readiness requires concrete structures. CIOs of Flexera add that the risk surface for AI has expanded so quickly that allocated handling time continues to rise. While a year ago the focus was on letting staff experiment, today autonomous agents can create other agents and discover vulnerabilities at unprecedented speed. Governance must constantly chase these changes. Leaders who overload themselves with AI governance are advised to prioritize the most critical controls before autonomous workflows exacerbate the problem.