Multiple outlets report that Palo Alto Networks’ State of Cloud Security Report 2026 finds many organizations have difficulty tracking what AI agents do during cloud and software development. The report also says AI-driven development is increasing the volume of infrastructure updates, while some organizations grant agents permissions that allow them to modify code or CI/CD configuration.

Both sources frame the issue as a security visibility and governance gap. Help Net Security highlights that a substantial share of organizations struggle to retrace AI agents’ actions, linking the challenge to release pressure and unresolved security concerns. SiliconANGLE emphasizes that, despite the widespread use of AI agents in development pipelines, many organizations cannot properly track or secure agent activities in the cloud. Together, the coverage points to difficulties in auditing autonomous or semi-autonomous workflows and managing the risks that come with moving agent-driven changes toward production.