#ML supply chain
Update: agent-framework attacker held a Microsoft release-pipeline key and ran an AI pentest tool, Wiz says
Wiz says the poisoned PyPI releases were pushed with a key for Microsoft's internal release service and three victims lost data. Microsoft hasn't confirmed. Keep versions pinned.
Microsoft's agent-framework package on PyPI was hijacked to steal Azure and CI credentials
Versions 1.26.1 and 1.26.2 of Microsoft Agent Framework for Python stole cloud and pipeline secrets on Nov. 17. Pin 1.26.0 and rotate secrets if you installed them.
Microsoft: about 1% poisoned training data can plant a hidden backdoor in AI coding agents
Microsoft researchers showed poisoned fine-tuning data or open-weight checkpoints can make coding agents run attacker commands on cue, while still passing benchmarks.
MLflow 3.18.0 fixes a flaw that let anyone write files on servers using the FileStore backend — update now
CVE-2026-53995 let unauthenticated users write files outside the trace folder on MLflow FileStore servers. The fix shipped in 3.18.0; the 3.17.2 patch release does not include it.
AWS warns of critical flaw in retired Multi Model Server; no patch will ship
CVE-2026-115812 (CVSS 9.8) lets a crafted model archive run code on AWS Multi Model Server 1.1.11 and earlier. MMS is end of life; AWS says migrate.
Fake "test" models on Hugging Face mined crypto on GPU servers — and a second wave followed the first takedown
Pluto Security found 31 Hugging Face repos whose custom code installs a crypto miner and backdoors when loaded with trust_remote_code. Downloads top 32,000; second-wave repos now disabled.
Two Unsloth flaws let a model's config file run code — one needs no trust_remote_code
A Pillar Security researcher found two ways a Hugging Face model's config.json could run code on machines using Unsloth. Both are fixed; only one got a CVE.