Gemini AI Hacked: Calendar Invite Used to Leak Private Data in Prompt Attack

• Researchers exploited Google's Gemini AI using a malicious calendar invite to leak private Google Calendar data.
• The attack, dubbed a 'prompt injection,' did not require technical sophistication, relying on basic language prompts.
• When a user with Gemini linked to their Google account analyzes the injected invite, the malicious payload activates.
• This vulnerability highlights how AI's ability to access personal data (Gmail, documents) can be weaponized.
• Google is actively developing solutions and encourages researchers to identify further vulnerabilities.

Why It Matters: Gemini AI was exploited via a simple calendar invite prompt injection, leaking private data and raising major security concerns.