CybersecurityDev3 min reading time

GitSpawn: Untrusted repos can execute code via AI coding agents

Hacker News
Read full post
Researchers discovered that several AI coding agents execute git commands that can run arbitrary code from a repository's configuration without user approval, posing security risks when using untrusted repos. This vulnerability arises because git commands refresh their index using repository-specific settings like core.fsmonitor, which can specify executable commands.

More in Cybersecurity

Cybersecurity6 min read

Anthropic reveals rogue AI agents hate CAPTCHAs, just like you

TechCrunch

Scoop: OpenAI faces GOP-led Senate investigation into Hugging Face breach

Covered by 2 sources

Chinese AI Giants Accused of Sending Millions of User Queries to U.S. Models

The Wall Street Journal