All news with #kubernetes security tag

🔬 This Google Cloud blog describes a three-layer reference architecture for building enterprise-grade, high-compute AI agents for drug discovery. It separates cognitive orchestration from GPU-bound execution and anchors both inside an enterprise governance plane. The post details execution choices (on-prem vs managed), containerized NVIDIA Inference Microservices on GKE, warm pools and preload strategies to avoid cold starts, and patterns for passing heavy payloads by reference. It also outlines Gemini platform governance features—agent identity, gateway, registry, and Model Armor—plus recommended asymmetric gating for reversible and irreversible actions.

🔒 Dell disclosed 18 CVEs affecting its Container Storage Modules (CSM) and Dell System Update (DSU), including two CVEs rated 10 and five rated 9+. The flaws could let unauthenticated or low-privilege attackers bypass auth, gain root, forge admin tokens, and execute remote code. Affected CSM versions prior to 1.17.0 and DSU versions prior to 2.3.0.0 must be updated; Dell reports no active exploitation so far. Vendors and customers are urged to apply fixes and rotate credentials.

🛠️ AWS Batch now supports Amazon EKS access entry authentication for compute environments. EKS access entries provide an API-driven way to grant IAM principals access to Kubernetes clusters, complementing the existing aws-auth ConfigMap. To enable, set accessEntry.desiredState to ENABLED via the CreateComputeEnvironment or UpdateComputeEnvironment APIs; AWS Batch creates one access entry per cluster and attaches the AWSBatchClusterPolicy. This feature is available in all Regions where AWS Batch is offered.

🔒 Dell patched multiple maximum-severity flaws in its Container Storage Modules (CSM) that link enterprise storage arrays to Kubernetes. The issues, rooted in the CSM Authorization module, allow unauthenticated attackers to obtain administrator credentials and bypass authorization. Dell recommends updating CSM to version 1.18.0 or later to remediate the critical vulnerabilities.

🚀 Google Cloud announces GKE Agent Sandbox and the Agent Sandbox RL orchestration SDK, now generally available, to address infrastructure bottlenecks in large-scale agentic reinforcement learning. The solution integrates SandboxWarmPool with GKE Image Streaming to eliminate cold-starts, reduce GPU idle time, and support thousands of large images with low TTFC. Native integrations with popular RL tools and an async Python SDK simplify orchestration and warm-pooling strategies for researchers.

🔍 This Unit 42 report examines how Kubernetes operators’ reliance on highly privileged service accounts creates a critical security weak spot, and introduces OperTraitor, an open-source LLM-powered engine that analyzes operator RBAC configurations. The tool compares documented functionality to granted privileges and assigns a normalized risk score, revealing abandoned or overly permissive operators in registries like OperatorHub. Case studies include a High-severity CVE in IBM’s Turbonomic and an overly permissive Datadog operator configuration, and the article offers practical mitigation guidance such as verifying sources, enforcing namespace-scoped operators, and continuously auditing RBAC.

🔥 Amazon EMR on EKS now supports interactive Apache Spark sessions via Spark Connect, enabling data engineers and scientists to develop and debug Spark applications from managed notebooks in Amazon SageMaker Unified Studio or their own IDEs like Jupyter and VS Code. An interactive session provides a persistent Spark context that spans cells and scripts, blending local Python execution with remote Spark operations on EKS. Sessions run as pods on virtual clusters secured by IAM execution roles and tagged by project and user, and Spark Connect is available in EMR release 7.14 (Spark 3.5) and emr-spark-8.1.0 (Spark 4.1) across AWS Commercial Regions.

🔍 A use-after-free bug in the Linux kernel's AF_UNIX socket garbage collector (CVE-2026-80521) can be abused to escape containers and gain host root, DepthFirst reported on September 22. The flaw was fixed upstream on August 6, but Ubuntu has not yet shipped patches for 26.04, 24.04, or 22.04 LTS; DepthFirst released exploit code targeting Ubuntu 26.04. The vulnerability is reachable from containers because AF_UNIX is allowed by default in common Docker and Kubernetes seccomp profiles, and no distro workaround has been published.

🚀 GKE Pod snapshots let you capture and restore a running Pod’s full state, including CPU and GPU memory, to dramatically reduce cold-start times for AI inference and sandboxed agent workloads. By persisting snapshots to high-throughput Cloud Storage, new replicas restore directly from a warmed state, cutting startup latency by up to 89% in benchmarks. This enables aggressive autoscaling, lowers idle GPU costs, and replaces complex custom caching solutions with a simple declarative CRD-driven workflow.

🎮 SeaVerse, a SeaArt gaming startup, built a platform for playable AI experiences and needed infrastructure that could run dynamic, multi-tenant sandboxes with low latency, strong isolation, and better observability. By adopting Google Kubernetes Engine (GKE) and GKE Agent Sandbox, SeaVerse achieved kernel-level isolation with microVMs and gVisor options, improved runtime visibility, and scaled sandbox allocations rapidly. The change reduced infrastructure costs by up to 60% while preserving a fast creator experience.

🚀 Filestore agent volumes deliver fully managed, high-performance elastic file storage tailored for large-scale agent fleets on Google Cloud. Integrated with Agent Substrate and GKE Agent Sandbox, volumes attach in milliseconds to provide isolated persistent workspaces with RWX support, POSIX semantics, and granular access controls. The feature targets non-production workloads now, with GA production access via allowlist.

🛡️ Agent Substrate is now available on Google Kubernetes Engine (GKE). This open-source agent execution runtime is engineered for high-density sandboxing, delivering sub-500ms resume times and hundreds of suspend/resume activations per second with native kernel and network isolation. Optimized for GKE but portable to any Kubernetes cluster, it supports hardware-isolated microVMs or gVisor sandboxes and integrates with existing agent frameworks like Hermes, Claude Code, and OpenClaw.

🔒 This Unit 42 report demonstrates how an attacker with root on a compromised Kubernetes node can abuse the SPIFFE/SPIRE machine identity system to impersonate co‑located workloads and harvest SVIDs. The research shows selector spoofing by manipulating cgroup and process metadata so the SPIRE agent issues another workload's identity. The authors provide an open‑source tool, Spooffe, to test and validate exposure and recommend hardening nodes, restricting root access, and minimizing privileged containers to reduce risk.

🔒 Amazon Elastic Kubernetes Service (Amazon EKS) now lets you associate up to 10 external OpenID Connect (OIDC) identity providers with a single cluster. This enables distinct user populations—employees, contractors, CI/CD systems—to authenticate directly without consolidating providers or using an identity broker. Each provider is configured and managed independently and coexists with existing IAM authentication. You add providers via the AWS Management Console or the AssociateIdentityProviderConfig API at no extra cost in all EKS regions.

🔒 ClusterNetworkPolicy (CNP) is a new cluster-scoped network policy API added to GKE to let administrators enforce deterministic, non-bypassable network guardrails across namespaces. CNP introduces a hierarchical tier model—admin, network policy, and baseline—with top-to-bottom evaluation and an explicit Pass action to delegate final decisions. Built with the Kubernetes SIG-Policy WG and implemented with Cilium, CNP is open source and intended to improve scalability, compliance, and portability of network security controls in multi-tenant clusters.

⚙️ Amazon EKS Provisioned Control Plane increases Horizontal Pod Autoscaler (HPA) sync concurrency up to 40× the default Kubernetes value, enabling faster scaling for HPA-driven workloads. This reduces the latency between increased demand and pod scale-out across clusters with many HPA objects. The change is applied for all Provisioned Control Plane customers with no configuration required. It enhances responsiveness for clusters operating at large scale.

🔒 This article presents a blueprint for securing AI workloads on Google Kubernetes Engine (GKE), consolidating controls across Google Cloud services and GKE features to create a secure-by-default platform. It covers three layers—infrastructure, supply chain, and application—and details capabilities such as Confidential GKE Nodes, Workload Identity Federation, k8s-aibom for AI SBOMs, Model Armor, and the GKE Inference Gateway. The blueprint recommends a three-phase rollout: Deploy, Operate, and Govern, and emphasizes integrating Google Cloud controls to maintain security at enterprise scale.

🛠️ This blog explains how to configure GKE Autopilot clusters to use GKE managed DRANET for GPU and TPU workloads. It outlines the setup flow: create a VPC, deploy an Autopilot cluster, define a custom ComputeClass, create a ResourceClaimTemplate for RDMA (GPUs) or netdev (TPUs), and deploy workloads that reference those resources. Examples and YAML snippets demonstrate GPU and TPU ComputeClasses, resource claim templates, and a deployment that binds pods to accelerators.

🔒 A critical vulnerability in Argo CD repo-server exposes risks inherent to GitOps platforms. Synacktiv found the unauthenticated GenerateManifest gRPC endpoint can be abused via Kustomize/Helm options to execute commands if an attacker can reach both the repo-server and Redis ports. The issue affects typical Helm deployments where Kubernetes network policies are not enabled by default, enabling lateral movement from a compromised pod. Synacktiv disclosed details July 1, 2026 and recommends strict network segmentation until a patch is available.

🛡️ AWS Network Firewall now supports container attribute-based rules for Amazon EKS and Amazon ECS, enabling firewall policies based on native container attributes (namespaces, pod names, labels, cluster names) instead of ephemeral IPs. This feature automatically discovers and tracks matching pods, dynamically updates IP-to-attribute mappings, and enriches alert logs with container context for easier troubleshooting and auditing. It integrates with Suricata-compatible rules, exports logs to CloudWatch Logs and S3, and can be configured via the AWS Console, CLI, or SDK with no additional feature charge.