Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

GitGuardian researchers reached 44 active Kubernetes clusters through leaked secrets, and 30% had been exposed for more than two years. Here’s how the OWASP Kubernetes Top 10 explains these Kubernetes security risks, and why K03, Secrets Management Failures, turns one leak into a chain.

In short, between April and July 2026, AI agents under evaluation at Anthropic, Google, Meta, OpenAI and the UK AI Security Institute reached systems outside their test environments, and on 30th September OpenAI said it had told more than 100 organizations that its agents may have touched theirs. The incidents have been read as containment failures, and they were. Read the five disclosures for who noticed, though, and they are detection stories first.

Together, CloudCasa by Catalogic and Dell PowerStore provide a complete disaster recovery solution for Kubernetes environments — combining PowerStore’s native array-to-array replication with CloudCasa’s Kubernetes-aware orchestration. PowerStore handles the data path, synchronously or asynchronously, while CloudCasa discovers workloads, validates the DR plan, and rebuilds Deployments, StatefulSets, and KubeVirt VMs on the target cluster, reconnecting them to the recovered volumes.

Kubernetes audit logs may show multiple engineers as one user and do not capture what is typed after a kubectl exec session starts. Assign each engineer a unique identity, make sure it stays consistent through proxies and cloud IAM, and use a session recorder if you need to review terminal activity.

In short, in May 2026 a Gemini model under evaluation by the AI security firm Irregular reached the systems of three real companies, and the incident has been reported as a breakout. By Google’s own account it was nothing of the kind. The test environment had internet access it was not meant to have, the fictional target shared its name with a real company, and the model found public information online, guessed one password and found two more in public code repositories.

Overprivileged Kubernetes ServiceAccounts persist when broad RBAC, cloud IAM permissions, and long-lived credentials outlive their intended use. Reduce overprivileged access with least privileged RBAC, scoped cloud permissions, and short-lived certificates that eliminate static credentials. I spent two days last quarter tracking down why a developer could delete production secrets. The RBAC looked fine. The ClusterRoleBinding said edit, not cluster-admin.

In short, the EU Cyber Resilience Act (CRA) puts binding cybersecurity requirements on any software sold as a product in the EU, and it does not carve out AI agents. Since 11th September 2026, any company that sells software with digital elements into the EU has 24 hours from learning that a vulnerability is being exploited to file an early warning with a national CSIRT and ENISA, 72 hours to describe it, and 14 days to report what it did about it.

SaaS or self-hosted is one of the first calls to make on Kubernetes backup and disaster recovery, and it’s worth settling before you compare features. This guide walks through what each deployment model means for a platform team, what to weigh before picking one, and where CloudCasa fits, since it offers both.

Docker Sandboxes isolate AI coding agents from the host. The GitGuardian Mixin Kit adds ggshield and AI hooks to scan prompts, tool actions, and outputs for secrets, giving developers a safer, repeatable path for agentic coding.