AITechTalks7 months ago
Prompt injection attack tricks Google's Antigravity into stealing your secrets
1 min read
A vulnerability in Google's Antigravity platform lets attackers steal credentials and code through indirect prompt injection. Security researchers PromptArmor hid malicious instructions in an online document. When a developer pointed Antigravity's AI agent to that URL, the agent bypassed security settings and exfiltrated sensitive data.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyGoogle
