February 2026 Summaries

Kubernetes posture management typically begins with basic guardrails, such as Pod Security Standards (PSS), which provide a foundation by blocking obviously unsafe workloads. However, these baselines can become inadequate as modern Kubernetes environments evolve to include complex workloads with unique security needs. The default Pod Security Admission (PSA) offers predictable but limited enforcement, focusing on broad security levels without considering specific workload contexts. As environments mature, the need for more expressive and contextual risk management becomes apparent, prompting some teams to explore more flexible admission frameworks like OPA Gatekeeper or Sysdig. These tools allow for more granular enforcement by considering workload identity, ownership, and other attributes, enabling risk-aware decisions that go beyond simple checklist compliance. While PSA establishes a baseline, tools like Sysdig can enhance posture management by incorporating vulnerability assessments and contextual considerations, allowing teams to maintain robust security without compromising operational needs. This shift from static checklists to informed decision-making ensures that Kubernetes security reflects the diverse and dynamic nature of modern workloads, balancing pre-admission controls with runtime security for comprehensive protection.

The blog post discusses the critical need for continuous trust in container security across the entire lifecycle, highlighting the challenges posed by the expanded attack surface due to container adoption. It explains how vulnerabilities can propagate through development, staging, and production environments, often undetected by traditional scanners, as illustrated by incidents like the Tesla Kubernetes breach. The partnership between CleanStart and Sysdig addresses these issues by bridging the gap between build-time and runtime security, using build-time hardening, cryptographic provenance, and deep runtime intelligence to create a continuous trust loop. This approach enables more effective vulnerability management by focusing on runtime risks and providing traceability from build to runtime. CleanStart's capabilities include delivering hardened container base images with minimal CVEs, aligning with SLSA principles, and offering custom image builds, while Sysdig extends trust into production with runtime threat detection and prevention using eBPF-based instrumentation and Falco rules. Together, they provide a comprehensive solution for container security, ensuring that organizations can prove their software's integrity and secure behavior in production.

LLMjacking, which emerged in May 2024 as a novel security threat, has evolved into a commercialized cybercrime marketplace by early 2026, paralleling the growth of cryptomining. This type of attack involves the unauthorized use of cloud-hosted Large Language Model (LLM) resources through compromised credentials, APIs, or exposed endpoints, leading to inflated cloud bills and potential exposure of sensitive model capabilities. Initially a theoretical concern, LLMjacking has developed into an organized ecosystem dubbed "Operation Bizarre Bazaar," where attackers monetize unauthorized AI access and sell it via underground marketplaces on platforms like Telegram and Discord. The risk to AI systems extends beyond financial costs to significant operational security threats, as attacks on Model Context Protocol (MCP) servers could lead to broader infrastructure compromises. For security leaders, this underscores the importance of robust credential management, an assume-breach mindset, and vigilant monitoring of APIs and AI integrations to mitigate the expanding risk landscape.

Cloud environments are becoming increasingly dynamic and are heavily influenced by AI, creating a gap between what security tools detect and what security teams can address, thus challenging Chief Information Security Officers (CISOs) with new operational risks. The widespread deployment of agentic AI and large language models (LLMs) generates additional noise and distractions, making deep runtime telemetry a critical priority to detect relevant threats. Existing security tools often miss new risk factors associated with AI, such as agent entitlements, decision-making, and communication, as well as model risks and protocol-related vulnerabilities. To effectively manage these risks, CISOs must focus on runtime insights that provide visibility into what is being executed and running in production, which is essential for ensuring observability, traceability, and explainability of AI systems. As enterprises generate value during runtime, prioritizing runtime telemetry helps filter the noise, sharpen priorities, and facilitate actionable security measures, making it indispensable for modern security architecture in AI and cloud-powered enterprises.

Sysdig has been recognized as a Leader in The Forrester Wave™: Cloud Native Application Protection Solutions, Q1 2026, highlighting its evolution from a container security specialist to a comprehensive CNAPP provider. The report, which evaluated 14 top CNAPP providers, underscores the importance of runtime protection in cloud security as organizations increasingly adopt AI-driven applications. Sysdig's platform, built on runtime-powered security, integrates posture, vulnerability, and runtime telemetry, providing a unified view of application risk. With AI-driven enhancements like Sysdig Sage™, the company aims to deliver real-time insights and protection, helping organizations prioritize and respond to risks effectively in dynamic cloud environments. The recognition by Forrester reflects Sysdig’s commitment to comprehensive cloud security and its ability to secure modern cloud environments at scale.

Running containers as unprivileged or rootless is highlighted as a crucial security practice to mitigate risks associated with running processes as root, which is the default setting in Docker and many other container runtimes. The article outlines how running containers with root privileges can expose systems to vulnerabilities, allowing attackers to gain full control if they manage to breach the container. It explains the steps to configure containers to run as unprivileged by adapting images, changing default ports, adjusting file permissions, and leveraging tools like User Namespaces and Capabilities to limit privileges. While making containers rootless can be complex and involves trade-offs, such as losing low-level access to certain resources, it significantly enhances security by isolating workloads and reducing potential damage from compromised containers. The article provides practical examples, such as modifying the nginx-unprivileged image, and suggests further steps to secure containers, including making binaries root-owned, using read-only modes, and employing multi-stage builds. Additionally, the use of Linux capabilities and user namespaces is advised for containers that require performing privileged operations, ensuring that even if a container is breached, the attacker has limited access to the host system.

In November 2025, the Sysdig Threat Research Team observed a rapid and sophisticated cyberattack on an AWS environment, where the attackers gained administrative access in under 10 minutes, leveraging large language models (LLMs) for automation. The attack began with the theft of credentials from public S3 buckets, which were then used for privilege escalation via Lambda code injection and lateral movement across 19 AWS principals. The attackers utilized Amazon Bedrock for LLMjacking, executed GPU instance provisioning for resource abuse, and employed a variety of techniques to evade detection, including IP rotation and role chaining. Sysdig's analysis highlighted the importance of employing the principle of least privilege and enhancing runtime detection to counteract such AI-assisted threats. The misuse of AI models, rapid enumeration of AWS services, and creation of backdoor access points underscore the evolving complexity of cloud-based cyber threats, with the attack demonstrating both the speed and potential for AI to significantly influence offensive operations in cloud security environments.

The text explores the security challenges and strategies associated with deploying AI applications on Oracle Kubernetes Engine (OKE) within Oracle Cloud Infrastructure (OCI), emphasizing the importance of a robust security posture for GPU-accelerated workloads. It highlights the shared responsibility model where Oracle manages the control plane while customers are responsible for application security and data-plane operations. The text identifies evolving threats in AI environments, such as model theft and data exposure, and underscores the need for runtime protection and real-time threat detection. Sysdig's approach to AI workload protection involves three pillars: runtime insights, agentic AI for threat response, and open innovation for transparency and control, complemented by integration with CI/CD and Kubernetes security posture management platforms. The piece also discusses the significance of starting with secure infrastructure blueprints and operationalizing security tools within existing stacks to address regulatory and organizational requirements effectively.

January 2026 began with significant cybersecurity challenges, as highlighted in a security briefing by Crystal Morin. The month saw multiple high-severity vulnerabilities and sophisticated malware threats, including the Ni8mare vulnerability impacting n8n servers, the VoidLink malware targeting cloud environments, and the ChainLeak vulnerabilities in the AI framework Chainlit. Sysdig's Threat Research Team provided in-depth analyses and mitigation strategies for these threats, emphasizing the increasing sophistication of attacks, such as the weaponization of GitHub Actions and the resurgence of LLMjacking. Additionally, the month featured reports of a Russian attack on Poland's energy grid and the release of a new AI compliance framework by the European Telecommunications Standards Institute. As threats continue to evolve, the briefing underscores the importance of vigilance and preparedness against the exploitation of open source software, advancements in cloud-native malware, and the integration of AI in attack strategies.