Four out of five enterprises fail to manage AI agent identities effectively

Four out of five enterprises that have implemented AI agent identities are struggling to manage those systems effectively, especially when faced with rogue incidents. This alarming statistic comes from recent research conducted by VentureBeat, which highlights the growing challenges organizations encounter as they integrate AI into their security frameworks.

The research indicates that more than half, or 53%, of enterprises have experienced a security incident or near-miss involving AI agents. Despite 65% of these organizations enforcing agent permissions during runtime, only 18% are isolating their highest-risk agents. Even more concerning, a mere 8% are combining both enforcement and isolation measures. This gap in security practices suggests that many enterprises are unprepared to handle the complexities of AI-driven security systems.

In a recent presentation at the VB Transform 2026 conference, Visa’s president of technology, Rajat Taneja, emphasized the importance of proactive security measures. He shared how Visa used Anthropic's AI model, Mythos, to identify vulnerabilities in its payment network. The company not only addressed these weaknesses but also open-sourced the framework that guided their security efforts. This level of engineering sophistication is not common among most enterprises, many of which rely heavily on native controls from AI platform providers, a practice that can deepen existing vulnerabilities.

The findings raise critical questions about the future of AI in enterprise security. As organizations increasingly rely on AI solutions, the ability to manage and contain these systems will be crucial. Without effective isolation and enforcement strategies, enterprises risk exposing themselves to serious security threats. The urgency to enhance agentic security measures is clear, as businesses navigate a rapidly evolving digital landscape fraught with potential risks.