Hackers could use poisoned WhatsApp and Slack notifications to take control of your Google Gemini and operate it on their behalf.


  • Fast injection flaw found in Android Gemini
  • Malicious notifications mix benign and hidden commands
  • Google fixed a server-side issue last November

Rapid injection attacks aren’t just for email messages or calendar entries. They can also be done on Android, using just about any communications platform in existence today. This is what SafeBreach researcher Or Yair said in a new report.

A prompt injection attack works by “injecting” a prompt where there should not be one. For example, an innocuous email might have a prompt hidden in white text on a white background, or written with font size 0, so the human can’t see it. However, if the victim asks their AI assistant to “read emails and sort them”, the assistant may treat the hidden text as a prompt and execute the wrong bids for the attackers.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top