AI adoption outpaces governance, leading to increased risk for enterprises

The rapid adoption of artificial intelligence (AI) by startups is creating a significant gap between the momentum of AI development and the associated risks, leading to what analysts are calling 'AI inflation' without a commensurate return on value, based on information from Arn Net.

Gartner reports that in 2025, only one in five AI projects are expected to yield a positive return on investment (ROI), a figure that remains unchanged a year later. While AI can increase productivity, rising costs often negate these gains, resulting in more ROI misses than hits. In Australia and New Zealand, 86% of CIOs report that risk is growing faster than value derived from AI initiatives. Vendors often overstate potential productivity increases, with Gartner data showing actual gains at only 16%. This push for AI consumption, even when unnecessary, leads to inefficiencies and waste.

For instance, using AI agents for simple tasks like title searches in a database is inefficient compared to function calls. Furthermore, AI generates waste, consuming resources and requiring rework. A recent study found that 40% of workers received AI-generated work that took nearly two hours to clean, costing large companies millions annually. This careless consumption diverts resources from critical AI use cases. The rapid pace of AI innovation, particularly with generative AI and agentic AI, compounds risks, including counterfeit reality, human displacement, wrong behavior, and potential irreversible harms.

Shadow AI, where AI is used without explicit IT approval, poses significant security risks, with incidents like the Hugging Face breach demonstrating the potential for AI agent swarms to compromise systems. Many AI vendors lack the stability, governance, and continuity expected by enterprises, creating a trust deficit. The rapid, uncontrolled pace of AI model innovation means that models can become outdated within months, breaking existing workflows and highlighting a lack of enterprise-readiness. Organizations must prioritize robust AI governance and cybersecurity frameworks to manage these escalating risks effectively.

Source: Arn Net