Researcher Details Critical Prompt-Injection Risk in ChatGPT GitHub Workflows
A researcher says hostile instructions hidden in a public repository could redirect an AI-assisted GitHub workflow toward a private repository and expose secrets. The reported 9.3-severity scenario highlights...
AgentCorruption Chain Showed How One Prompt Could Expose AWS AgentCore Secrets
Zenity researchers demonstrated how a malicious prompt could turn an AgentCore agent’s legitimate tools and cloud role into a route to credentials, conversations, code, and secrets. AWS has...
Encrypted Prompt Injection Can Turn Copilot CLI Into a Developer-Secret Exfiltration Tool
Researchers demonstrated how encrypted instructions on a hostile web page could lead Copilot CLI in autopilot mode to read local files and transmit secrets. The finding shows why...
Russia-Aligned Group Tests Prompt Injection to Blind AI Malware Scanners
A Russia-aligned campaign used a malicious script comment designed to trigger an AI model’s safety refusal and interrupt malware analysis. The GuardBreaker technique, found in a MATCHBOIL delivery...
AI Gateways Under Fire: Attackers Chain LiteLLM and MCP Flaws for Remote Code Execution
A 90-day honeypot study shows attackers systematically probing AI proxies, MCP servers, and agent frameworks like LangChain and Langflow, chaining authentication bypass and command-injection bugs into remote code...
One Malicious Webpage Can Hijack Your AI Coding Agent Through an NVIDIA NemoClaw Flaw
A critical flaw in NVIDIA's NemoClaw tooling exposes a local AI inference server to the open network, letting a single malicious website hijack an AI agent via DNS...
Grok AI Chatbot Tricked Into Leaking Private Chats Through Encrypted Prompt Injection
Security researchers at Adversa AI found a zero-click flaw in xAI's Grok that hides malicious instructions inside encrypted text to steal names, locations, and chat history. The attack...
Researchers Find Matching RCE Flaws in Claude Code, Gemini CLI and Codex Coding Agents
Security researcher Elad Meged has uncovered a strikingly similar vulnerability pattern across AI coding agents from Anthropic, Google, and OpenAI, all traceable to how each vendor's surrounding 'harness'...