Two compromised versions of LiteLLM sat on PyPI for roughly 40 minutes on the morning of March 24, 2026. That window was enough to capture secrets from around 434,000 CI/CD pipeline runs across nearly 2,500 organizations, including AWS, Samsung, Cisco, Salesforce, Siemens, and Deloitte. In August, researchers at CloudSEK and Hudson Rock confirmed they had obtained the raw exfiltrated data: a 153GB archive containing 433,909 files of environment variables, cloud keys, Kubernetes secrets, and API tokens harvested live from running pipelines, as covered by Help Net Security’s reporting on the credential archive. If LiteLLM runs anywhere in your stack, or you touch any AI proxy infrastructure at all, you need answers to three things: whether you were exposed, what to rotate first, and whether the rotation you did back in March actually held. That last one matters more than it sounds, because “we rotated everything” has already burned at least one very large company. How the Breach Happened The attack didn’t start with LiteLLM. On March 19, 2026, a threat group called TeamPCP compromised the build pipeline of Trivy, a vulnerability scanner half the industry runs, and pushed a poisoned release. LiteLLM’s own CI pipeline ran Trivy, so the poisoned scanner had legitimate read access to the project’s runner environment. The attackers used that to steal LiteLLM’s PyPI publishing tokens and ship two malicious releases of their own: versions 1.82.7 and 1.82.8. KICS and the Telnyx Python SDK got hit in the same campaign. The payload design is the part worth studying. The malicious package dropped a .pth startup hook into site-packages, so the code ran the moment any Python interpreter started on the machine, whether or not anything imported LiteLLM. From there it harvested environment variables, read local credential files like .aws/credentials and .kube/config, tried to move laterally across Kubernetes clusters, and installed a systemd backdoor dressed up as a generic telemetry service. InfoQ’s coverage of the PyPI compromise put downloads of the compromised release above 40,000. For scale, LiteLLM normally gets downloaded around 3 million times a day. The exfiltration had a nasty fallback, too. According to CloudSEK, stolen data was encrypted and sent to a typosquatted domain, and when that failed, the malware created a public repository inside the victim’s own GitHub account and uploaded the loot as a release asset. Some companies were publishing their own secrets to the open internet and had no idea. Worth Knowing: The malicious code only existed in the PyPI artifacts. The GitHub source repository stayed clean the whole time, so a developer reviewing the code on GitHub saw nothing wrong. Source review isn’t artifact verification. If you don’t check that what the registry serves matches the upstream source, this class of attack is invisible to you. How to Check If You Were Exposed Three checks, from quickest to most involved. 1. Confirm whether the compromised versions ever ran The malicious versions went live on PyPI at 10:39 UTC on March 24, 2026 and got quarantined about 40 minutes later. The project’s advice: treat any install from that day before 16:00 UTC as suspect. Search your lockfiles, pip caches, SBOMs, and container image histories for 1.82.7 and 1.82.8. And check your internal artifact mirrors. An Artifactory or Nexus proxy that cached the bad release in March can keep serving it internally long after PyPI pulled it. Keep the .pth mechanism in mind when you scope this. The question isn’t “which applications import LiteLLM,” it’s “which machines had the package installed at all,” because every Python process on an infected machine triggered the payload. 2. Hunt for persistence Rotation is pointless if the attacker still has a foothold. Check developer machines, CI runners, and containers for unauthorized .pth files in site-packages and for suspicious systemd units, especially anything posing as a system telemetry service. And review activity from March 24 onward, not just the 40-minute window. Persistence is there so the access outlives the infection. Pro Tip: Don’t limit the persistence hunt to live machines. Base container images rebuilt in late March may have baked the payload into every image derived from them since. Scan your image registry for the affected LiteLLM versions and for unexpected .pth files, then trace which running workloads came from flagged images. 3. Check whether your secrets are in the dump Hudson Rock has published a domain lookup tool and is running ethical disclosures for affected organizations, and CloudSEK maintains a high-confidence victim list. Use them, but know their limits. Attribution in this dataset is genuinely hard. One dump with a siriusxm.com committer email actually traced, through its self-hosted GitLab endpoints, to AdsWizz, a SiriusXM subsidiary. And a large share of the dumps are generic pipeline configurations with no identifying domain, email, or server name at all. Absence from a victim list is not evidence of absence. If your pipelines ran the compromised versions, assume exposure no matter what a lookup tool tells you. What to Rotate, in What Order The guidance from both research teams is blunt: treat every secret the LiteLLM environment could reach as compromised. That covers secrets on disk, in memory, injected into CI jobs, and anything retrievable through instance metadata services. Work down by blast radius: Priority Credential type Why it comes first 1 Cloud IAM keys (AWS, GCP, Azure) Direct control of infrastructure, data stores, and billing. This is where attackers monetize fastest. 2 GitHub and GitLab PATs, package publishing tokens These let an attacker poison your releases and turn your company into the next link in the supply chain. 3 Kubernetes service account tokens and kubeconfigs Lateral movement across clusters was built into the payload, not a theoretical risk. 4 Database passwords and third-party API keys Dumped in plain text in the archive, often with no attribution, so nobody will warn you they leaked. 5 AI provider API keys Billing abuse, quota theft, and access to whatever data flows through your LLM routing layer. One word matters more than the rest of this article: revoke, don’t just rotate. That
Most organizations get ISO 42001 certified in 2 to 9 months. Companies that already hold ISO 27001 regularly land in the 2 to 5 month range, while enterprises with sprawling AI portfolios and no existing management system can take 12 months or more. The audit itself only takes days. Almost the entire calendar goes into building and operating your AI Management System (AIMS) long enough to produce evidence an auditor can actually check. That is the short answer. The longer answer depends on your starting point, your scope, and how quickly you can get a certification body on the schedule. This article breaks down the full timeline phase by phase, the factors that stretch or compress it, and what the recertification cycle looks like once you hold the certificate. Typical ISO 42001 Certification Timeline at a Glance ISO/IEC 42001:2023 is the first international standard for AI management systems, published in December 2023. Because it follows the same harmonized structure as ISO 27001 and ISO 9001, the certification process will feel familiar to anyone who has been through a management system audit: build the system, run it, pass a Stage 1 and Stage 2 audit, then maintain it through annual surveillance. Here is how timelines typically break down by company size. Average Timeline for Small Businesses Small companies move fastest because scope stays contained. A startup with two or three AI systems, a handful of decision makers, and short approval chains can finish scoping in a week and get policies signed off in days rather than weeks. The realistic floor for a small business starting from scratch is around 3 months. With an existing ISO 27001 program and a compliance platform already collecting evidence, 2 months is achievable. Average Timeline for Mid-Sized Companies Mid-sized companies usually take 6 to 9 months. The AI inventory is growing, more departments are touching AI systems, and risk assessments have to cover more use cases. Coordination becomes the hidden cost: getting engineering, legal, and product to agree on an AI policy takes longer than writing the policy itself. Average Timeline for Enterprises Enterprises should plan for 9 to 12 months, sometimes longer. The main drivers are AI system sprawl across business units, longer procurement cycles for certification bodies, and audits that take more days. The Stage 2 audit for a large multinational can run two weeks or more on its own, and internal alignment before the audit takes far longer than the audit itself. Breakdown of the ISO 42001 Certification Timeline by Phase The phases below overlap in practice. Treat the durations as effort estimates for a reasonably resourced program, not a strict sequence. Phase 1: Scoping and Gap Analysis (2–4 Weeks) Everything starts with two questions: which AI systems are in scope, and how far is your current governance from what the standard requires? The gap analysis maps your existing policies and controls against the standard’s clauses and Annex A controls, and produces the project plan for everything that follows. Get the scope wrong here and every later phase inherits the mistake. Phase 2: AIMS Design, Leadership, and AI Policy Development (2–4 Weeks) This phase establishes the skeleton of the management system: the AI policy, governance roles, objectives, and the leadership commitments the standard requires. Executive sign-off is the gating item. The documents are not hard to write. Getting senior leadership to formally own AI governance is where programs stall. Phase 3: AI Risk and Impact Assessments (2–6 Weeks) ISO 42001 requires both AI risk assessments and AI impact assessments, and the distinction matters. Risk assessments look at what could go wrong for the organization. Impact assessments look at consequences for individuals and society, which is a newer discipline for most teams. This phase takes longer when you have many AI systems, high-risk use cases, or no prior methodology to adapt. The output feeds directly into your Statement of Applicability (SoA), the document that maps which Annex A controls you have selected and why. Insider Note: Impact assessments are where auditors probe hardest, because they are the most distinctive part of ISO 42001 compared with ISO 27001. A recycled security risk register with “AI” pasted into it will get picked apart in Stage 2. Build the impact assessment methodology properly the first time. Phase 4: Controls Implementation (2–10 Weeks) The longest phase. Here you implement the Annex A controls selected in your SoA: AI system lifecycle documentation, data governance for training data, human oversight mechanisms, transparency measures, supplier management for third-party AI, and so on. Duration depends almost entirely on the gap analysis results. Organizations with mature engineering practices often find they already do much of this and just need to document it. Organizations without formal AI development processes are building from zero. Phase 5: Documentation, Training, and Evidence Collection (2–8 Weeks) Certification requires proof that the system operates, not just that it exists on paper. That means records: training completion logs, risk assessment outputs, review meeting minutes, monitoring reports. This phase runs partly in parallel with implementation, but it cannot be compressed below a certain floor because auditors want to see evidence generated over time, not a folder of documents all created the week before Stage 1. Phase 6: Internal Audit and Management Review (2–4 Weeks) The standard requires an internal audit of the AIMS and a formal management review before the certification audit. This is your dress rehearsal. A good internal audit surfaces nonconformities while they are still cheap to fix. Skipping or rushing it is a false economy that shows up later as Stage 2 findings. Phase 7: Stage 1 Certification Audit (1–2 Weeks) The certification body reviews your documentation and assesses readiness for Stage 2. The audit itself takes 1 to 3 days for most organizations. The auditor examines your scope statement, AI policy, risk and impact assessment methodology, SoA, and internal audit results, then issues findings. The 1–2 week window covers the audit plus the report. Phase 8: Closing Nonconformities (2–4 Weeks) Almost every Stage 1 produces findings.
