An IBM study of 2,000 CIOs and CTOs shows two-thirds are responsible for AI systems they can’t fully oversee, with 77% saying AI adoption is outpacing their governance frameworks. Organizations that build controls into their AI deployments report fewer incidents, higher margins and can scale agent use far more effectively than those relying on manual oversight.

• Two-thirds of CIOs/CTOs are accountable for AI systems they can't fully oversee, and only 11% feel ready for large-scale agent deployment despite expecting 38% more agents by 2027.
• Companies logged 54 AI agent incidents on average last year, with 17% high-severity and most causing data breaches, cascading failures, or compliance violations.
• Building governance controls directly into AI deployments (vs. manual oversight) cuts incidents by 25%, enables 16x more agent deployment, and boosts operating margins by 18%.
• 84% of organizations haven't operationalized AI budgeting and 85% lack real-time spend visibility, yet disciplined firms deploy 2.4x more agents without increasing budgets.