Tag: Claude Code
Plugin4Shell breaks plugin pinning across four AI coding agents
A shared design weakness in the major coding agents lets hostile plugins run with the same reach as the developer…
Poisoned Git configs make AI coding agents run attacker commands
Manifold finds eight flaws where poisoned Git settings make seven AI coding agents run attacker commands.
Poisoned logs turn trusted platforms into agent hijackers
Tenet Security's Ghostjacking demo shows attackers planting instructions in logs that AI agents read and execute, targeting Cloudflare, Datadog, and…
Chinese actors use Claude Code and DeepSeek to automate cyber espionage
Security researchers uncovered state-sponsored operators using Claude Code and DeepSeek models to automate intelligence gathering and maintain persistent access to…
AI Tools Claude and GPT 4.1 Used to Coordinate Mass Government Data Breach in Mexico
The attacker processed data from 305 servers and generated 2,597 intelligence reports using a custom Python script linked to OpenAI's…