CVE-2026-100843: MONAI: unsafe pickle deserialization enables RCE
HIGHPoC AVAILABLECISA: ATTENDMONAI versions before 1.6.0 deserialize files with pickle.loads() in algo_from_pickle() (monai/auto3dseg/utils.py), so a crafted pickle file runs arbitrary commands as the user who loads it. Exploitation needs local access and a user to open the file (CVSS 7.8, EPSS 0.00196, top 91% percentile), and there is no public exploit, no Nuclei template and no CISA KEV listing, though CISA SSVC rates it ATTEND. The blast radius is moderate: 126 downstream dependents, 14 other CVEs in the same package, and a medical imaging user base where hosts often hold sensitive patient data and GPU clusters. Upgrade to MONAI 1.6.0 or later, and until then never load Auto3DSeg pickle artifacts from untrusted or unverified sources.
What is the risk?
Medium-high. Impact is full compromise (C/H, I/H, A/H) of the host running the code, but exploitation needs user interaction and a local attack vector, which limits mass exploitation. Exploitation likelihood is low in absolute terms (EPSS 0.2%) but sits in the top 91% percentile. Risk rises sharply where researchers share Auto3DSeg algorithm artifacts across teams, download them from public hubs, or run pipelines on shared GPU servers with access to PHI and model IP. Unsafe deserialization (CWE-502) is a well-understood class that attackers can weaponize easily once a delivery path exists.
How does the attack unfold?
What systems are affected?
| Package | Ecosystem | Vulnerable Range | Patched |
|---|---|---|---|
| MONAI | pip | — | No patch |
Do you use MONAI? You're affected.
How severe is it?
What is the attack surface?
What should I do?
1 step1) Upgrade MONAI to 1.6.0 or later (pip install -U monai) and rebuild container images. 2) Until patched, avoid calling algo_from_pickle() on any file that is not generated and stored by your own trusted pipeline. 3) Restrict write access to directories and buckets that hold Auto3DSeg artifacts, and verify integrity with checksums or signatures. 4) Scan pickle artifacts with tools such as picklescan or fickling before loading, and prefer non-executable serialization formats. 5) Run training jobs with least privilege, in sandboxed containers with no standing cloud credentials and restricted egress. 6) Detect: use SCA or pip-audit to inventory MONAI versions below 1.6.0, and alert on unexpected child processes spawned by Python training jobs.
What does CISA's SSVC say?
Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.
How is it classified?
Which compliance frameworks are affected?
This CVE is relevant to:
Frequently Asked Questions
What is CVE-2026-100843?
MONAI versions before 1.6.0 deserialize files with pickle.loads() in algo_from_pickle() (monai/auto3dseg/utils.py), so a crafted pickle file runs arbitrary commands as the user who loads it. Exploitation needs local access and a user to open the file (CVSS 7.8, EPSS 0.00196, top 91% percentile), and there is no public exploit, no Nuclei template and no CISA KEV listing, though CISA SSVC rates it ATTEND. The blast radius is moderate: 126 downstream dependents, 14 other CVEs in the same package, and a medical imaging user base where hosts often hold sensitive patient data and GPU clusters. Upgrade to MONAI 1.6.0 or later, and until then never load Auto3DSeg pickle artifacts from untrusted or unverified sources.
Is CVE-2026-100843 actively exploited?
Proof-of-concept exploit code is publicly available for CVE-2026-100843, increasing the risk of exploitation.
How to fix CVE-2026-100843?
1) Upgrade MONAI to 1.6.0 or later (pip install -U monai) and rebuild container images. 2) Until patched, avoid calling algo_from_pickle() on any file that is not generated and stored by your own trusted pipeline. 3) Restrict write access to directories and buckets that hold Auto3DSeg artifacts, and verify integrity with checksums or signatures. 4) Scan pickle artifacts with tools such as picklescan or fickling before loading, and prefer non-executable serialization formats. 5) Run training jobs with least privilege, in sandboxed containers with no standing cloud credentials and restricted egress. 6) Detect: use SCA or pip-audit to inventory MONAI versions below 1.6.0, and alert on unexpected child processes spawned by Python training jobs.
What systems are affected by CVE-2026-100843?
This vulnerability affects the following AI/ML architecture patterns: training pipelines, AutoML pipelines, medical imaging ML workflows, shared model and artifact repositories, research notebooks.
What is the CVSS score for CVE-2026-100843?
CVE-2026-100843 has a CVSS v3.1 base score of 7.8 (HIGH). The EPSS exploitation probability is 0.30%.
What is the AI security impact?
Affected AI Architectures
MITRE ATLAS Techniques
AML.T0010.003ModelAML.T0011.000Unsafe AI ArtifactsAML.T0018.002Embed MalwareAML.T0050Command and Scripting InterpreterCompliance Controls Affected
What are the technical details?
Original Advisory
MONAI versions before 1.6.0 contain a remote code execution vulnerability in the algo_from_pickle() function due to unsafe pickle.loads() deserialization in monai/auto3dseg/utils.py. Attackers can craft malicious pickle files that execute arbitrary system commands when deserialized by the vulnerable function.
Exploitation Scenario
An attacker publishes a pre-trained Auto3DSeg algorithm package, or tampers with one in a shared research repository or storage bucket, embedding a malicious pickle that runs a reverse shell on load. A data scientist downloads the artifact and restores it with algo_from_pickle() on a GPU workstation or shared training server. The payload executes with the user's privileges, steals cloud and registry credentials, exfiltrates imaging datasets and model weights, and gives the attacker a foothold to move laterally into the hospital or research network.
Weaknesses (CWE)
CWE-502 — Deserialization of Untrusted Data: The product deserializes untrusted data without sufficiently ensuring that the resulting data will be valid.
- [Architecture and Design, Implementation] If available, use the signing/sealing features of the programming language to assure that deserialized data has not been tainted. For example, a hash-based message authentication code (HMAC) could be used to ensure that data has not been modified.
- [Implementation] When deserializing data, populate a new object rather than just deserializing. The result is that the data flows through safe input validation and that the functions are safe.
Source: MITRE CWE corpus.
CVSS Vector
CVSS:3.1/AV:L/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:HReferences
Timeline
Related Vulnerabilities
CVE-2026-487059.3MONAI: NaN/inf DICOM input causes DoS and corrupt inference
Same package: monaiCVE-2025-587558.8MONAI: path traversal allows arbitrary file write
Same package: monaiCVE-2025-587568.8MONAI: unsafe deserialization in CheckpointLoader allows RCE
Same package: monaiCVE-2025-587578.8MONAI: unsafe pickle deserialization RCE in data pipeline
Same package: monaiCVE-2026-1008428.8MONAI: eval sandbox bypass allows code execution
Same package: monai