tag: Cloud-Credentials · 4 items
- Engineer — Act: Exposed credentials are confirmed valid and immediately usable by anyone. Audit your org’s public Gitee repositories for hardcoded cloud and AI API keys, then rotate any discovered secrets and enforce pre-commit secret scanning across all public code-hosting platforms.
- SOC/IR — Plan: No specific IOCs or active campaign are reported, but the scale of valid exposed credentials signals likely ongoing abuse. Build or tune detections for anomalous cloud API authentication — unusual geos, times, or access patterns for cloud/AI service keys — to catch credential misuse before it surfaces as a breach.
- Leader — Plan: At 26,400 working credentials, this research underscores systemic secret sprawl risk beyond GitHub. This quarter, confirm whether developer activity extends to Gitee or similar platforms, and verify that automated secrets scanning is enforced on all public repositories your org controls.
- Engineer — Act: Active mass-scanning campaign targeting internet-exposed Vite dev servers to exfiltrate AWS/Azure credentials and infrastructure state files; audit for any Vite dev servers reachable from the internet, restrict network access immediately, and rotate cloud credentials for any instances that may have been exposed.
- SOC/IR — Plan: No specific IOCs surfaced in available reporting, but this active campaign warrants building detections for anomalous AWS/Azure API calls (unusual regions, new IAM key usage, credential enumeration) that could indicate stolen dev-server credentials being leveraged.
- Leader — Plan: Cloud credential theft from exposed dev tooling can cascade into full infrastructure compromise; confirm with engineering that no Vite dev servers are internet-facing and that credential rotation procedures cover this scenario before it becomes a material incident.
- Engineer — Act: MLflow is common in cloud-hosted ML pipelines and the SSRF flaw enables IMDS credential theft — active exploitation corroborated by two independent sources (watchTowr, VulnCheck). Patch MLflow to the fixed version immediately and audit IMDS endpoint access controls on any host running it.
- SOC/IR — Act: Active exploitation of MLflow SSRF is confirmed by two independent sources, with cloud credential theft as the objective. Hunt for anomalous outbound requests to IMDS (169.254.169.254) originating from ML pipeline hosts, and check for SSRF-pattern HTTP requests against MLflow endpoints since early August.
- Leader — Plan: If your org runs MLflow in cloud environments, active exploitation of this SSRF flaw creates cloud credential-theft risk for data science or AI teams. Confirm engineering has inventoried and patched MLflow deployments this sprint and review whether any cloud credentials may have been exposed.
- Engineer — Act: Actively scanning for internet-exposed instances of ComfyUI, Ollama, n8n, Open WebUI, Langflow, and Gradio to harvest AWS keys and Kubernetes tokens — exactly the stack teams deploy fast without firewall controls. Audit now for public exposure of these service ports, restrict to internal networks, and rotate AWS/K8s credentials on any host that ran them exposed.
- SOC/IR — Plan: The TTPs are concrete enough to build detections around: Shodan-driven scanning targeting AI service endpoints, followed by credential exfiltration. Build hunts for unusual outbound traffic or credential API calls originating from AI service hosts; the summary appears truncated so IOCs are not yet available to act on directly.
- Leader — Plan: A claimed harvest of 3,811 AWS keys illustrates the systemic risk of teams rapidly standing up AI infrastructure without security review. Raise with engineering and DevSecOps leadership to establish a deployment standard for AI tooling that includes network isolation requirements before services go live.