Azure AI Foundry: AI Development Platform
Azure AI Foundry enables creation of AI agents that run multi-step jobs, which can be exploited via MCP tool description poisoning.
Tag: AI Agent
Microsoft Purview Data Loss Prevention
Purview DLP helps prevent data exfiltration by monitoring and controlling data transfers, relevant for AI agent attacks.
Microsoft Entra Agent ID for Identity Management
Entra Agent ID provides identity management for AI agents, helping to enforce least privilege and monitor actions.
Microsoft Defender for Cloud Security
Defender for Cloud offers security monitoring and threat protection for cloud environments, including AI agent activities.
Microsoft Sentinel SIEM Solution
Sentinel is a cloud-native SIEM that can log and analyze AI agent actions to detect anomalies like data exfiltration.
Microsoft Warns Poisoned MCP Tool Descriptions Can Make AI Agents Leak Data
Microsoft has issued a warning about a new attack vector targeting AI agents that use the Model Context Protocol (MCP). Attackers can…
Invariant Labs Demonstrates Tool Poisoning Attack on AI Agents
Invariant Labs first named 'tool poisoning' in April 2025, showing how hidden instructions in tool descriptions can hijack AI agents like Cursor…
Copilot Studio: Custom AI Agent Builder
Copilot Studio allows building custom AI agents that can reach into business systems, increasing the attack surface for poisoned tool descriptions.
Plandex AI Agent Vulnerable to GuardFall Shell Injection Attack
Adversa AI demonstrated a full end-to-end attack against the production Plandex binary, exploiting the GuardFall bypass to execute malicious shell commands.
Hermes Project First to Document GuardFall Shell Injection Bug
The Hermes project was where the GuardFall vulnerability first surfaced, documented in its issue tracker, affecting its AI coding agent.