12 posts
A Go-based propagation system ("n4d" / "Mesh Control") targets internet-exposed MCP (Model Context Protocol) servers that offer code-execution tools without authentication. Enrolled scan nodes receive CIDR assignments from a centralized C2 panel at 209.99.186.235, probe every IP…
The article details a precision prompt injection attack methodology against AI agents, using a fictional travel agent called 'Varda' as a case study. The attack follows a kill chain approach: reconnaissance to extract system prompt logic, enumerate tools, and learn data schemas; then weaponization to craft a fake payment confirmation that satisfies the agent's preconditions for booking flights. The core vulnerability is that the LLM treats conversation history as trusted context, allowing attackers to inject fabricated tool responses and fake assistant messages that bypass sequential validation checks, enabling unauthorized action execution without proper authorization.
The upcoming MCP 2026-07-28 specification fundamentally reshapes the protocol's security model by moving to a stateless architecture, eliminating protocol-managed sessions, and mandating OAuth 2.1 with PKCE. While this removes historical attack vectors like session hijacking and unsolicited server prompts, it introduces new risks: client-controlled state objects and tracking IDs enable cross-agent workflow hijacking, the _meta object allows metadata-based privilege escalation, new HTTP headers create desync and data leakage opportunities, MCP Apps bring stored XSS into AI interfaces, and asynchronous tasks introduce resource exhaustion DoS vectors. Security responsibility now rests squarely on MCP server developers and platform operators to implement cryptographic state verification, input validation, output encoding, and resource quotas.
A coordinated supply chain attack compromised 19 PyPI packages, utilizing malicious .pth files to achieve execution at Python startup. The loader downloads the Bun runtime to execute an obfuscated JavaScript stealer targeting developer secrets, cloud credentials, and CI/CD tokens, exfiltrating data via GitHub repositories and Actions.
Security researchers discovered critical vulnerabilities in three widely used Model Context Protocol (MCP) servers—Apache Doris, Apache Pinot, and Alibaba RDS—stemming from insufficient back-end security validation. These flaws include SQL injection (CVE-2025-66335), missing authentication, and unauthenticated data exposure, allowing attackers to execute arbitrary commands or exfiltrate sensitive database metadata.
AI agents deployed in enterprise environments are highly susceptible to indirect prompt injection attacks, enabling data theft and unauthorized actions. Security teams must adopt an 'assume breach' architecture for LLMs, focusing on blast radius reduction through agent sandboxing, credential isolation, egress restrictions, and human-in-the-loop governance.
Model Context Protocol (MCP) servers introduce a new attack surface akin to AI-native APIs, exposing organizations to protocol-level attacks, injection vulnerabilities, and authorization bypasses. Because MCP tools often use permissive validation to accommodate LLM inputs and proactively broadcast their capabilities via plain-English descriptions, attackers can easily map business logic and exploit downstream systems or trigger resource exhaustion.
Microsoft has released the open-source Agent Governance Toolkit to address the growing security risks associated with autonomous AI agents. The toolkit provides runtime policy enforcement, cryptographic identity, and execution sandboxing to mitigate threats outlined in the OWASP Top 10 for Agentic Applications, though challenges in credential scoping and semantic intent classification remain.
The article outlines the emerging security risks associated with autonomous Agentic AI and presents a collaborative architectural solution between TrendAI and NVIDIA. By integrating TrendAI's governance and behavioral analysis with NVIDIA's OpenShell runtime, enterprises can safely deploy self-evolving AI agents with runtime policy enforcement and protection against AI-native threats like prompt injection.
TrendAI researchers demonstrated novel attack vectors against AI systems, including exploiting AI-driven KYC pipelines using 'executable documents' to leak customer data. Additionally, they introduced FENRIR, an automated vulnerability hunting system that has discovered numerous zero-days in AI and Model Context Protocol (MCP) ecosystems.
Check Point Research discovered critical vulnerabilities in Anthropic's Claude Code CLI that enable Remote Code Execution (RCE) and API token exfiltration. By injecting malicious configurations into project files like .claude/settings.json and .mcp.json, attackers could execute arbitrary commands and steal API keys when a developer opens a compromised repository, leading to potential supply chain attacks and unauthorized access to shared Claude Workspaces.
AI Rush Opens New Attack Paths as Trusted Cloud Services Fuel Phishing The rush to adopt artificial intelligence is giving attackers two new advantages: convincing lures to trick users and poorly secured infrastructure to exploit. This week, multiple campaigns used fake websites for the Claude AI assistant to infect victims with password-stealing malware, while researchers revealed that commercial robots and AI connection protocols contain critical flaws that let hackers hijack them. Because organizations are deploying AI tools faster than they can secure them, attackers are finding easy entry points into corporate networks. In parallel, phishing campaigns are increasingly hijacking trusted cloud services like Amazon's email platform and Vercel's AI-powered website builder to send messages that bypass security filters entirely. A massive campaign targeting US employees used fake HR reviews to steal login sessions even when multi-factor authentication was enabled, and the breach of the Canvas learning platform exposed data on 275 million people that can now be used for highly convincing follow-up scams. These trends together suggest that traditional defenses are losing effectiveness because attackers are hiding inside the systems we already trust. Organizations should immediately patch the actively exploited Palo Alto Networks and Ivanti vulnerabilities flagged by CISA this week, require phishing-resistant authentication methods, and treat every AI tool and robot connected to their network as a high-risk device that needs strict monitoring.
