Updating the taxonomy of failure modes in agentic AI systems: What a year of red teaming taught us
2026-06-05T14:52:16Z•78a702fe0066b2b9a926347f8741df3a1d3428cb9d461ac47ad566312da8e3cd
.netagentic-aici-cdconfluencecredential-theftcryptojackingdependency-confusiondetection-and-mitigationf5-big-ipkerberos-relaylateral-movementnpmransomwarered-teamingscreenconnectself-propagationsoftware-supply-chainsupply-chaintyposquatting
What happened
A set of Microsoft Security Blog posts describing multiple high-impact threats and mitigation guidance: updated taxonomy of failure modes for agentic AI after a year of red‑teaming; several large npm supply‑chain campaigns (Red Hat “Miasma”, dependency‑confusion and typosquatting campaigns) that steal GitHub, cloud and CI/CD credentials and propagate by republishing packages; an analysis of The Gentlemen self‑propagating Go ransomware; a cryptojacking campaign abusing SEO poisoning, ScreenConnect and .NET utilities for GPU mining; and a multi‑stage Linux intrusion that leveraged an exposed F5
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- microsoft_security_blog
- Record identifier
- 78a702fe0066b2b9a926347f8741df3a1d3428cb9d461ac47ad566312da8e3cd
- Enrichment time
- 2026-06-05T14:52:16Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.