A malicious version of Anthropic's official Claude extension for the Cursor IDE was compromised for 12 hours, potentially affecting up to 5,000 developers. The backdoor aimed to install persistent SSH backdoors and steal cloud credentials and API tokens. Neither Anthropic nor Cursor have yet explained how the attackers gained access. AI
IMPACT Highlights supply chain risks for AI developer tools and the potential for widespread credential theft.
RANK_REASON A security vulnerability in a third-party integration of an AI model's extension.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →