Executive Summary
PointGuard AI's incident tracker lists AI security issues including Microsoft Semantic Kernel flaws enabling prompt injection to host-level RCE and CrewAI issues involving prompt injection, code execution and arbitrary file access.
What Happened
PointGuard AI’s incident tracker lists vulnerabilities involving prompt injection, chatbot guardrail bypass, data exposure, and agent workflow compromise. Its entries include Microsoft Semantic Kernel flaws reportedly enabling prompt injection leading to host-level remote code execution, as well as CrewAI issues involving prompt injection, code execution, and arbitrary file access.
Affected Products
- Microsoft Semantic Kernel
- CrewAI
Vulnerability / Attack Technique
Prompt injection escalating to code execution in agent frameworks
Technical Mechanism
In the listed framework flaws, injected instructions reach code-execution or file-access functions, giving host-level control.
AI Security Category
LLM01:2025 Prompt Injection
Part of the Prompt Injection topic.
Relevant Frameworks
RealGround Analysis
Agent frameworks turn prompt injection into code execution when model output reaches powerful functions. RealGround analysis: framework choice and version are security decisions.
Recommended Controls
- Patch Semantic Kernel and CrewAI to fixed versions
- Disable code-execution tools unless required
- Validate model output before passing it to file or shell functions
Related RealGround Service
AI Agent Business Logic Audit · or start with the free AI Agent Security Scanner.
Sources
PointGuard AI: https://www.pointguardai.com/ai-security-incident-tracker
Summary and analysis written by RealGround from the source above; last updated 2026-10-17.