UTC

Mistral AI SDK compromise

Part of the May 2026 Mini Shai-Hulud wave. A compromised developer device led to malicious Mistral SDK releases on PyPI (mistralai 2.4.6, which has no matching tag or pipeline run) and npm. The PyPI release fetched a credential harvester on Linux; Mistral describes the npm payload as non-functional.

First seen
2026-05-11
Disclosed
2026-05-12
Affected packages
4
Attributed to
No public attribution

What to do

Remove the affected versions from systems, lockfiles, caches, images and mirrors; mistralai 2.4.5 and earlier are not affected and current 3.x releases are clean. Rotate secrets reachable from any system that imported 2.4.6 and review cloud audit logs, as Mistral's advisory MAI-2026-002 describes.

Check a lockfile for these releases →
Affected releases
PackageRegistryVersions
mistralaiPyPI2.4.6
@mistralai/mistralainpm2.2.2, 2.2.3, 2.2.4
@mistralai/mistralai-azurenpm1.7.1, 1.7.2, 1.7.3
@mistralai/mistralai-gcpnpm1.7.1, 1.7.2, 1.7.3
Sources
  1. https://docs.mistral.ai/resources/security-advisories/MAI-2026-002 · Mistral AI, 2026-05-12
  2. https://github.com/mistralai/client-python/security/advisories/GHSA-wx9m-wx4f-4cmg · Mistral AI / GitHub Advisory Database, 2026-05-18
  3. https://github.com/advisories/GHSA-jgg6-4rpr-wfh7 · GitHub Advisory Database, 2026-05-18
  4. https://github.com/mistralai/client-python/issues/523 · mistralai/client-python (GitHub), 2026-05-12

Written up from the sources above and last checked on 2026-11-18. ShieldMap describes what was affected and what to do; it does not publish the malicious code or how it worked. Report a correction to [email protected].