Cyber Risks · The Hacker News

Google Gemini Prompt Injection Flaw Exposed Private Calendar Data via Malicious Invites

Summary: Miggo Security researchers disclosed a Google Gemini flaw in which a malicious calendar invite carried a dormant natural-language prompt that, once a user asked Gemini an ordinary scheduling question, caused the chatbot to summarize the user's private meetings into a new calendar event. Miggo's Liad Eliyahu said the bypass allowed unauthorized access to private meeting data and creation of deceptive calendar events without direct user interaction, and the issue was addressed after responsible disclosure.

Bottom line: In many enterprise calendar configurations the attacker-created event was visible to the attacker, so private meeting data left the organization without the target user taking any action at all.

Read the original at The Hacker News ↗

Topics
AI Security
Related