LiteLLM Supply Chain Poisoning Exposes 195TB of Credentials Across 2500 Organizations
Security researchers have disclosed the full scope of the LiteLLM supply-chain poisoning incident that occurred in March, describing it as one of the largest AI-related credential leaks to date. Within roughly 40 minutes, attackers exfiltrated approximately 195TB of credentials belonging to around 2500 organizations, including major technology companies such as Nvidia.
LiteLLM serves as a widely used open-source API gateway for connecting Python applications to services from OpenAI, Anthropic, and Azure. The project records nearly 100 million monthly downloads and is installed approximately 3.4 million times per day, making it a critical piece of AI infrastructure.
Attackers from the group TeamPCP did not target the LiteLLM repository directly. Instead, they compromised a dependency used inside LiteLLM’s CI/CD pipeline: the open-source vulnerability scanner Trivy. By rewriting a Git tag, they injected malicious code into Trivy version v0.69.4. Because LiteLLM’s pipeline pulled the latest Trivy tag without version pinning, the poisoned scanner executed during the build process and stole PyPI publishing tokens.
Using the stolen tokens, the attackers published two malicious LiteLLM releases—1.82.7 and 1.82.8—on PyPI. These packages were downloaded more than 119,000 times before PyPI isolated them. The malware collected and transmitted cloud provider keys, Git tokens, SSH private keys, Kubernetes configurations, environment variables, and API keys for AI services.
Analysis by CloudSEK and Hudson Rock identified roughly 434,000 CI/CD-related credentials within the 195TB dataset. The first user report to PyPI arrived only after one hour and nineteen minutes, underscoring the lack of automated detection for malicious package behavior. A coding flaw in the malware’s .pth launcher triggered recursive fork bombs that crashed victim machines, inadvertently limiting the duration of silent operation.
Security analysts emphasize that modern software supply chains rely on transitive trust across hundreds or thousands of dependencies. In this case, the attack surface extended beyond LiteLLM’s own code to every third-party tool referenced in its build pipeline.
Recommended mitigations include strict version pinning with exact hashes in requirements files, locking GitHub Actions to specific commit hashes rather than moving tags, immediate rotation of any credentials used between March 24 and the discovery date, and deployment of environment-variable exfiltration monitoring inside CI/CD workflows.
The incident demonstrates that AI tooling and infrastructure components are now attractive targets for supply-chain adversaries. Organizations are advised to treat AI dependencies with the same rigor applied to production code and to prepare for an increase in similar attacks against model-serving frameworks, vector databases, and training pipelines.
Related articles
Protestware: How Ideologically Motivated Attackers Compromise Open Source Code and Supply Chains
Protestware represents a growing supply chain risk where open source maintainers deliberately insert destructive or politically motivated code into widely used packages. Unlike traditional malware seeking financial gain, protestware is driven by ideological motives and can delete files, cause denial of service, or restrict functionality based on geography or locale. Notable incidents include the node-ipc and peacenotwar packages that overwrote files on systems in Russia and Belarus, the colors.js infinite loop causing DoS, and es5-ext triggering high CPU usage during specific hours or Russian locales. These attacks often hide in postinstall hooks, use geolocation checks via ipinfo.io, or activate via environment variables and timestamps. Detection requires static analysis for geo-IP logic, dynamic monitoring in sandboxes, and tools such as Software Bill of Materials and OpenSSF Scorecard. Organizations are advised to adopt Zero Trust for dependencies, pin versions, use internal artifact repositories, and follow frameworks including NIST SSDF and SLSA to mitigate risks.
Attackers Exploit Critical JFrog Artifactory Flaw CVE-2026-82329 to Forge Admin Tokens and Poison CI/CD Pipelines
A critical vulnerability tracked as CVE-2026-82329 in JFrog Artifactory is being actively exploited in the wild, allowing unauthenticated attackers with network access to generate or forge administrator tokens on self-managed installations that retain default settings. The flaw enables full authentication bypass, granting attackers administrative privileges to enumerate users and groups, access sensitive configurations, read stored artifacts, and modify security settings. The greatest risk arises when Artifactory serves as the central repository for automated CI/CD pipelines, where an attacker can replace trusted internal packages, container images, or dependencies and propagate malicious changes downstream into build runners and production environments. JFrog has released patches for multiple branches, including versions 7.111.21, 7.117.28, 7.125.20, 7.133.29, 7.146.38, and 7.161.20, although patching alone does not invalidate already-issued tokens. Organizations are advised to restrict network exposure, revoke and rotate admin tokens, review audit logs for anomalous token generation and configuration changes, and validate artifact integrity across the vulnerable period. The issue does not affect JFrog Cloud instances.
Critical JFrog Artifactory Flaw CVE-2026-82329 Actively Exploited to Forge Admin Tokens in Self-Managed Deployments
A critical vulnerability identified as CVE-2026-82329 in JFrog Artifactory is being actively exploited in the wild, enabling unauthenticated attackers to generate or forge administrator tokens simply by having network access to affected instances. The flaw impacts self-managed installations that retain default configurations, a common scenario when repositories are deployed quickly and later exposed to the internet. Once administrative access is obtained, attackers can enumerate users and groups, access sensitive configurations, read stored artifacts, and modify security settings. The most severe risk arises when Artifactory automatically feeds CI/CD pipelines, allowing malicious substitution of trusted packages, container images, or dependencies that then propagate downstream into production environments. JFrog Cloud instances were already protected, while self-managed deployments require updates to specific patched versions including 7.111.21, 7.117.28, 7.125.20, 7.133.29, 7.146.38, or 7.161.20. Even after patching, previously issued tokens must be revoked because their validity operates independently of the software update. Organizations are advised to restrict network exposure, rotate credentials, audit token generation logs, and validate artifact integrity for any material published during the vulnerable period.
Unicode Tricks Let Malicious Python Code Bypass Code Review
A detailed analysis shows how subtle Unicode manipulations allow code to pass human review while executing entirely different logic. The first technique replaces Latin characters with visually identical Cyrillic or other script letters inside identifiers, such as using Ukrainian 'і' instead of Latin 'i' in an is_admin variable. Python treats these as distinct names because NFKC normalization does not map Cyrillic to Latin. The second technique inserts bidirectional override characters that reorder text for the compiler while the editor displays the intended order. The third hides zero-width characters inside string literals to break comparisons and searches. A compact Python auditor using the tokenize module and explicit character sets detects all three classes of characters. The article recommends adding such checks to CI pipelines and configuring linters including ruff and flake8 to reject suspicious commits.