Shadow Leak Zero-Click Flaw Leaks Gmail Data via OpenAI ChatGPT Deep Research Agent
Summary: Radware researchers disclosed a zero-click flaw, codenamed ShadowLeak, in OpenAI ChatGPT's Deep Research agent that lets an attacker leak Gmail inbox data with a single crafted email, using indirect prompt injection hidden in email HTML through tiny fonts, white-on-white text, and layout tricks. OpenAI addressed the issue in early August following responsible disclosure on June 18, 2025; the attack requires the Gmail integration to be enabled but can extend to any supported connector, including Box, Dropbox, GitHub, Google Drive, HubSpot, Microsoft Outlook, Notion, and SharePoint.
Why it matters: The exfiltration happens inside OpenAI's cloud environment, not on the endpoint, so local and enterprise defenses never see it - and the same technique reaches every connector ChatGPT supports, not just Gmail.
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