News

AI Governance Gap Leaves Enterprises Struggling To Spot Problems Before Intervening

Human oversight may be a cornerstone of enterprise AI governance, but it offers limited protection when organizations cannot reliably detect problems that require intervention.

New research from Tata Consultancy Services (TCS) suggests that businesses have made considerably more progress establishing controls over AI decision-making than developing the real-time monitoring capabilities needed to enforce them.

The TCS Enterprise AI Study, based on responses from 1,333 senior business and technology leaders across 25 countries and 12 industries, highlights the disparity. While roughly two-thirds of respondents said their organizations prohibit autonomous AI decisions or confine them to limited use cases, just 9 percent rated their real-time AI performance monitoring as highly and consistently effective.

The findings raise questions about how well companies can exercise meaningful human oversight as AI becomes more deeply embedded in business operations. By comparison, 32 percent gave the same rating to their ability to manage exceptions, escalations, and human intervention.

Human Oversight Remains Central to AI Decisions
According to the research, approximately two-thirds of respondents said their organizations either do not allow AI to execute decisions without human approval or permit autonomous decisions only in a few specific use cases.

The findings suggest that while businesses are increasingly using AI to support decision-making, they remain cautious about allowing systems to act independently.

However, the study identified a gap between companies' ability to intervene and their ability to detect when intervention is necessary.

Security, risk and governance considerations also ranked as the leading factors influencing where and how organizations develop and deploy AI, ahead of cost, speed and ease of deployment.

AI Governance Remains Fragmented
The research team found that 42 percent of organizations either manage AI governance separately across platforms and teams or rely primarily on individual vendors' basic governance capabilities.

Featured