Tag
#AI Coding Agents
AI Coding Agents Expose Sensitive Internal Images on GitHub
A recent discovery by Glow, a security company, reveals that AI coding agents have inadvertently exposed over 13,000 internal images from more than 300 organizations on public GitHub repositories. These images include sensitive information such as customer billing records and screenshots of unreleased features. The exposure occurred when developers were asked to share screenshots of code changes for review, highlighting a previously unconsidered risk in the use of AI tools for coding assistance. Organizations are advised to review their use of AI coding agents and implement stricter controls on image sharing.
Zero-Click RCE Flaw in AI Coding Agents: Plugin4Shell
A zero-click remote code execution (RCE) flaw, dubbed Plugin4Shell, was discovered in popular AI coding agents such as OpenAI's Codex, Anthropic's Claude Code, Google's Gemini CLI, and Microsoft-owned GitHub Copilot. This vulnerability allowed attackers to execute malicious code without developer interaction by swapping a trusted plugin with a malicious one, potentially gaining a foothold in enterprise development environments. Researchers at AIR reported the flaw to the vendors, and most have released patches. Users must update their agents to mitigate the risk.