Signal is active with 2 mentions in latest observed window
Recommended action window: Monitor and triage in normal cycle
NVD description
In MONAI 1.6.0, PersistentDataset (monai/data/dataset.py) explicitly rejects the combination track_meta=True with weights_only=True, forcing users who cache MetaTensors (the default tensor type in MONAI >= 1.0) to run torch.load(hashfile, weights_only=False). Related cache helpers in monai/data/utils.py also call pickle.loads on cached content and derive cache keys with hashlib.md5. As a result, a local user with write access to a shared or world-writable cache_dir (e.g. /tmp/monai_cache, HPC scratch, ~/.cache/monai) can place a malicious pickle file that is deserialized the next time another user's MONAI pipeline reads the cache, resulting in arbitrary code execution in that user's context. All released versions of the monai pip package are affected; no patched version is available as of the advisory.
Priority
LOW
Exploitation
NONE
PoC
NONE
Patch
NONE
Momentum
NONE
To cache MetaTensors, MONAI's error message tells you to switch off PyTorch's weights_only guard. Today's CVE says no fix exists; the fix shipped nine days ago in a release candidate. No exploitation reported. https://severitydaily.com/monai-cve-2026-100841-patched-none-fixed-1-6-1rc0-release-candidate/
CVE-2026-100841: high-severity flaw in MONAI (all versions through 1.6.0). • PersistentDataset forces unsafe torch.load on cached MetaTensors • Local user who can write to a shared cache can run code • CVSS 8.5, no stable patch yet https://thecircuitry.to/article/monai-160-hit-by-high-severity-rce-flaw-via-pickle-cache-muk13d8j