An IBM survey of 2,000 C-level executives found that only 11% feel fully prepared for AI agent deployment, highlighting a growing control gap as AI use expands across businesses. The survey, conducted in 2026, revealed that two-thirds of CIOs and CTOs said they were accountable for AI systems they did not fully control, while 70% said teams were deploying technology faster than IT could track.

Madhuri Chandoor, founder of PromptHalo, an AI security and trust infrastructure company, emphasizes the importance of distinguishing capability from authority in AI agent deployment. She argues that organizations need to consider the context and downstream impact of an action before an AI agent proceeds. Chandoor recommends behavioral profiling for AI agents, similar to financial fraud monitoring, to identify potential risks and ensure that agents operate within predetermined limits.

Chandoor illustrates the concern with a hypothetical example of an infrastructure-managing AI agent that autonomously adds or removes an index or changes table structures in a production environment, potentially affecting live transactions, customer data, or dependent processes. She also uses a refund scenario to demonstrate how an AI agent may issue refunds up to $50 without human review, but a user could request ten $50 refunds rather than one $500 refund requiring review, highlighting the need for broader session context and behavior review.

Chandoor's approach involves documenting the resources an AI agent can access, the conditions that apply to that access, and the possible downstream effects of particular actions. She also recommends observability gates to review activity, particularly to inspect and address when requests become repeated, unusually broad, or inconsistent with the purpose originally assigned to the agent.

By adopting a more nuanced approach to AI agent deployment, businesses can ensure that they are using AI responsibly and verifying its behavior throughout the process. As Chandoor emphasizes, establishing clear accountability ownership across organizations for AI applications security is essential to operationalizing these guardrails and pursuing AI innovation while giving security and accountability the necessary attention.

This article was written with the assistance of AI.
News Factory APP - agentic news to boost your SEO & AEO.