Security · 1h ago
AI Coding Tools Face Trust Crisis: Grok Build Uploads Repos, Claude Code Leaks Source
Security researcher cereblab found Grok Build silently uploading entire codebases to Google Cloud, ignoring the 'Improve the model' toggle. Anthropic's Claude Code leaked source code three times in 13 months, with China's MIIT flagging it for covert surveillance. GitHub Copilot and Cursor also suffered security incidents, highlighting systemic failures in AI coding tools.
Meridian48 take
The industry's 'move fast and break things' approach is breaking developer trust—these aren't isolated bugs but a pattern of negligence that demands regulation.
Read the full reporting
Grok Walks Off With Your Repo? When the "Smartest Assistant" Becomes the Most Dangerous Tool →
DEV Community
ai-coding-toolsdata-privacy