Google Patches Antigravity IDE Flaw Enabling Prompt Injection Code Execution
Summary: Researchers at Pillar Security found a flaw in Google's Antigravity agentic IDE that combined its permitted file-creation ability with insufficient input sanitization in the native find_by_name tool, letting an attacker inject the -X (exec-batch) flag through the Pattern parameter so fd would run arbitrary binaries against workspace files. The tool call executed before Strict Mode's network, write, and sandbox constraints were enforced, and Google patched the issue as of February 28 following a January 7, 2026 disclosure.
What you should know: The attack needs no account compromise: a user pulling a file with hidden instructions from an untrusted source is enough, because the trust model assuming a human will catch something suspicious does not hold when autonomous agents follow instructions from external content.
Read the original at The Hacker News ↗
- Topics
- AI Security
- Related
- SECURITY AFFAIRS AI-CYBERSECURITY NEWSLETTER ROUND 8 Cyber Risks · Nov 15, 2026Pentagon Renames CDAO ‘Super Intelligence Office’ Cyber Risks · Nov 15, 2026186 Fake AI Domains Redirect Users to Click Fix Lures, Browser Malware, and Search Hijackers Cyber Risks · Nov 12, 2026