Enterprise AI Agent Trust Grows Despite Widespread Outages and Governance Gaps

Enterprise AI Agent Trust Grows Despite Widespread Outages and Governance Gaps

Companies are placing unprecedented trust in AI agents to handle critical operations, yet many lack the proper safeguards to manage these powerful systems effectively. A new survey from PagerDuty reveals both the promise and peril of rapid AI adoption in the enterprise.

Trust Meets Reality: A Dangerous Gap

The survey of 1,500 IT and business executives across six countries shows 81% are now willing to let AI agents take autonomous action during crises like security breaches. Meanwhile, 74% of companies consider AI so essential they would struggle without it. This sounds impressive until you consider the operational reality: 84% have already experienced AI-related outages.

Even more concerning is the confidence gap. While 96% of executives believe they can detect and prevent AI failures, the widespread outage data suggests otherwise. This overconfidence becomes especially risky when companies are betting their operations on systems they may not fully understand.

Key Concerns for Business Leaders

Testing Lags Behind Adoption

• 84% use AI to write or review code
• Only 39% consistently test AI-generated code
• US companies lead testing practices (59%) while Japanese organizations lag (19%)

Governance Is an Afterthought

• 85% acknowledge needing better error detection procedures
• Companies with multiple AI agents are more likely (79%) to worry about exceeding management capabilities
• 25% have deployed five or more AI agents without adequate oversight frameworks

Recovery Plans Are Inconsistent

• When outages occur, 52% complete tasks manually
• Smaller companies (under 10,000 employees) are better prepared with response plans (63%)
• Separate Capgemini research shows trust in AI agents actually declined from 43% to 27% among experienced organizations

What This Means for Your Business

The rapid pace of AI deployment may be creating more risk than reward. Companies are scaling AI faster than they can develop proper governance, testing, and incident response capabilities. Before expanding AI initiatives, consider whether your organization has the operational foundations to support autonomous systems safely.

The data suggests a "walk before you run" approach may be wiser than rushing to deploy multiple AI agents without adequate oversight.

🔗 Read the full analysis on Diginomica