Agent Security

Agent security focuses on constraining, observing, and recovering AI agents that can use tools, credentials, cloud resources, or production workflows.

Key facts

  • Cloud SDK flaws and storage-path assumptions matter more as agents automate model upload, deployment, and data-pipeline work [src-181].
  • AI Watch treats agent security as an operational discipline spanning credentials, workflow mutation, audit logs, and rollback.

Related

Evidence checked on 14 September 2026

Unit 42 reports a human-directed intrusion using AI agents, more than 50 ATT&CK techniques and less than ten hours. Mandatory branch protection blocked attempted infrastructure-code backdoors. [src-441] One responder account, partly relying on attacker statements; the corrected report describes intrusion and extortion rather than ransomware deployment.

Robin Cartier perspective

This page is part of Robin Cartier's working AI knowledge graph: a practical research layer for production AI, recommendation systems, experimentation, GEO, and agentic web readiness.

The useful next step is to connect this concept back to applied product leadership and operating models.

Recommended next

Keep reading from this thread

From 500 indexed pages and articles.

  1. Wiki concept Vertex AI Historical source material recovered on 2026-09-09. Capture dates and source limitations apply; recovery is not fresh verification. Related by security
  2. Wiki concept AI Cybersecurity Governance Covers the policies and operational controls needed to secure AI systems, including model supply chains, cloud SDKs, agent permissions, storage Related by security
  3. Wiki concept AI Supply Chain Security Historical source material recovered on 2026-09-09. Capture dates and source limitations apply; recovery is not fresh verification. Related by security