Security company Glow found over 13,000 internal images exposed through public GitHub repositories by AI coding agents, including billing records and unreleased features.
Mandiant reveals how an attacker hijacked an AI coding assistant session at a software provider and used it to spread Shai-Hulud across around 100 internal repositories.
Researchers have disclosed GhostSplice, a prompt injection technique that enables malicious MCP servers to manipulate AI coding agents into exposing sensitive data through fragmented instructions.
North Korean Contagious Interview linked PolinRider campaign publishes 108 malicious packages across npm, Go, Packagist and Chrome extensions targeting developers with BeaverTail and RAT payloads through compromised repositories and VS Code automation.
Security researchers identified 73 fake Microsoft VS Code extensions on Open VSX tied to GlassWorm v2 malware campaign targeting developers through supply chain attacks and sleeper packages.
A supply chain attack linked to Trivy has led to the spread of CanisterWorm malware across dozens of npm packages, exploiting tokens and decentralized infrastructure.
Security researchers report an expanded GlassWorm campaign using malicious Open VSX extensions and hidden Unicode code to target developers and steal sensitive data.