Security researchers have identified five serious vulnerabilities in Fluent Bit, the lightweight and widely used open-source observability agent. These issues, reported by Oligo Security, can be chained to bypass authentication, manipulate tags, perform path traversal, and even achieve remote code execution. When exploited, an attacker could disrupt cloud services, overwrite files, tamper with logs, or crash the agent. The affected components include tag-processing logic, the Docker Metrics input plugin, and the in_forward protocol, exposing organisations to risks such as file corruption, spoofed telemetry, and denial-of-service conditions. The vulnerabilities, tracked as CVE-2025-12969, CVE-2025-12970, CVE-2025-12972, CVE-2025-12977, and CVE-2025-12978, enable a wide range of attacks. These include crafting containers with extremely long names to trigger buffer overflows, injecting traversal sequences through malformed tags, and spoofing trusted tag patterns to reroute or falsify logs. CERT/CC noted that an attacker with network access to a vulnerable Fluent Bit instance could exploit these flaws to execute malicious code, inject fabricated events, or hide evidence by rewriting log entries—potentially gaining a deeper foothold into cloud or Kubernetes environments. Fluent Bit maintainers have patched the issues in versions 4.1.1 and 4.0.12, with AWS urging customers to upgrade immediately. Additional defensive steps include avoiding dynamic tagging for routing, restricting output path expansion, mounting configuration files as read-only, and running Fluent Bit under non-privileged accounts. These measures help prevent runtime tampering and mitigate risks to cloud infrastructures that rely heavily on Fluent Bit for log collection and processing.
A critical security issue has been identified in Google Cloud’s Vertex AI platform that allows low-privileged users to escalate privileges and compromise high-permission service ...
A set of critical vulnerabilities has been identified in CrewAI, a widely used platform for building multi-agent AI systems. These flaws expose environments to prompt injection att...
A critical security flaw in Oracle WebLogic Server has rapidly become a prime target for attackers worldwide. Identified as CVE-2026-21962, the issue carries the highest possible s...