What’s new: Researchers from Tel-Aviv University have demonstrated a method using Google’s Gemini AI that could potentially allow hackers to take control of smart home devices. This method relies on “promptware” techniques, which can manipulate Gemini to execute unauthorized actions without the user’s knowledge. Google has acknowledged the issue and implemented multiple defenses to mitigate these risks.
Who’s affected
Users of smart home devices integrated with Google services may be at risk if vulnerabilities in AI-powered assistants are exploited through indirect prompt injection techniques.
What to do
- Ensure that all smart home devices and associated applications are updated to the latest versions to benefit from security patches.
- Monitor account activity for any unauthorized actions or changes.
- Report any vulnerabilities or suspicious activity to Google through their Bug Hunters program.
Sources
