https://www.msbiro.net/tags/prompt-injection/ Recent content in Prompt-Injection on Cloud Native & Open Source: A Team Lead’s Working Journal https://www.msbiro.net/%3Clink%20or%20path%20of%20image%20for%20opengraph,%20twitter-cards%3E https://www.msbiro.net/%3Clink%20or%20path%20of%20image%20for%20opengraph,%20twitter-cards%3E Hugo -- 0.164.0 en-us Wed, 05 Aug 2026 07:00:00 +0000 https://www.msbiro.net/posts/owasp-genai-llm-top-10-2026/ Wed, 05 Aug 2026 07:00:00 +0000 https://www.msbiro.net/posts/owasp-genai-llm-top-10-2026/ OWASP released the GenAI LLM Top 10 2026, the first edition grounded in 7,714 real AI security incidents. Prompt Injection stays at number one. Excessive Agency climbs. Misinformation is the widest gap between what practitioners fear and what the evidence shows. https://www.msbiro.net/posts/shadow-ai-cicd-kubernetes-threat-model/ Mon, 27 Jul 2026 00:00:00 +0000 https://www.msbiro.net/posts/shadow-ai-cicd-kubernetes-threat-model/ Shadow AI turns ungoverned coding assistants and agents into a live threat across CI/CD. A stage-by-stage threat model from developer laptop to Kubernetes pod, with the executive controls and tooling that contain the risk. https://www.msbiro.net/posts/zero-trust-for-ai-agents-anthropic-ebook/ Wed, 10 Jun 2026 06:42:27 +0100 https://www.msbiro.net/posts/zero-trust-for-ai-agents-anthropic-ebook/ A review of Anthropic's Zero Trust for AI Agents eBook: a practical security framework for deploying autonomous AI agents, covering threat modeling for security leaders and an implementation guide for architects and engineers.