70% of Enterprises Lack Stage-Three AI Agent Security Controls, VentureBeat Survey Finds

Quick Facts

• 70% of enterprises have not completed the transition to stage-three AI security controls that isolate agent execution
• 88% of organizations reported confirmed or suspected AI agent security incidents in the last year
• Shadow AI breaches cost an average of $4.63 million per incident, $670,000 more than standard breaches

Most enterprises cannot stop advanced AI agent threats despite widespread deployment across their organizations, according to a new VentureBeat survey. The study reveals a critical gap between executive confidence and actual security controls.

The survey found that 82% of executives report confidence that existing policies protect against unauthorized agent actions. But the data tells a different story. Unauthorized tool or data access ranked as the most feared failure mode, growing from 42% in January to 50% in March.

VentureBeat identified a three-stage AI security framework. Stage one focuses on observation. Stage two adds enforcement through identity and access management integration. Stage three implements isolation with sandboxed execution to limit damage when safeguards fail.

No provider ships a complete stage-three security stack today. Most enterprises remain stuck at observation while their agents already need isolation controls.

The speed of threats compounds the problem. CrowdStrike’s Falcon sensors detect more than 1,800 distinct AI applications across enterprise endpoints. The fastest recorded adversary breakout time has dropped to 27 seconds. Monitoring dashboards built for human-speed workflows cannot keep pace with machine-speed threats.

According to IBM’s 2025 Cost of a Data Breach Report, shadow AI breaches cost an average of $4.63 million per incident. Healthcare organizations face the highest incident rates at 92.7%.

The average organization now manages 37 deployed agents. That number grows every quarter as individual teams spin up automation without central review. A 2026 Gravitee survey found that only 24.4% of organizations have full visibility into which AI agents communicate with each other.

Policy enforcement consistency grew from 39.5% to 46% between January and February, the largest gain of any security capability. But monitoring investment patterns show enterprises struggling with the transition. Investment snapped back to 45% of security budgets in March after dropping to 24% in February.

Security leaders recognize the scale of the challenge. A Dark Reading poll found that 48% of cybersecurity professionals identify agentic AI and autonomous systems as the most dangerous attack vector. Another survey showed 92% are concerned about AI agents’ impact on workforce security.

The government has taken notice. Recent guidance defines AI agent systems as consisting of generative AI models with scaffolding software that can take discretionary actions with little human oversight. Officials warn that security vulnerabilities may pose future risks to critical infrastructure.

Read more: Most enterprises can’t stop stage-three AI agent threats, VentureBeat survey finds