State of AI infrastructure report agent governance and security
A new report highlights security and governance challenges as major barriers to scaling AI agents in enterprises, with 79% of tech leaders citing these as critical concerns.
AI agents, which can read emails, query databases, and trigger actions, pose new security risks alongside productivity benefits. A survey found 79% of tech leaders view security, governance, or operations as their biggest challenge in scaling inference. Autonomous workflows have altered enterprise risk models, introducing threats like tool poisoning and indirect prompt injection, where attackers manipulate agent logic through processed data. Legacy security tools struggle to address these dynamic threats, particularly in managing agent permissions and multi-system access.
To mitigate risks, organizations are adopting integrated cloud platforms with full-stack oversight. According to the report, 69% of executives now consider a full-stack platform critical, with 80% prioritizing data compliance in their choice. Frameworks like the Secure AI Framework (SAIF) and platforms such as Gemini Enterprise Agent Platform are being used to centralize control. These approaches focus on secure-by-default design, agent governance, and human-in-the-loop approvals to manage risks effectively.
Security leaders are shifting from breach prevention to verifying data provenance to counter misuse, including indirect prompt injection. The report emphasizes that simply locking down systems is counterproductive for autonomous agents. Instead, organizations are embedding security into AI development processes and adopting purpose-built tools for agent permission and identity management. These measures aim to expose blind spots and limit risks associated with agent interactions.
The report argues that robust governance within a unified foundation enables organizations to deploy AI agents confidently across critical workloads. By re-architecting their security stack, leaders in the agentic era are using security as a foundation for innovation. This approach allows them to scale faster while maintaining control over sensitive operations, positioning governance as a driver for secure and scalable AI adoption.