ZDNET’s Key Takeaways
Researchers demonstrated a way to hack Google Home devices via Gemini, a major AI model. This raises concerns about the safety of smart home technology and the potential for malicious control. Google responded by putting additional safeguards in place.
What’s the concern?
The fear that AI can be used to control our lives is already a barrier for many to adopt the new technology. Now, the possibility of smart devices being hacked adds to the anxiety. What if someone could use AI to manipulate our smart home devices, such as turning on a boiler or opening shutters?
- Artificial intelligence (AI) has the potential to be used for malicious purposes.
- Smart home devices can be vulnerable to hacking.
How the attack worked
Researchers from multiple institutions, including Tel Aviv University, Technion, and SafeBreach, conducted a controlled, indirect prompt injection attack. They embedded malicious instructions into Google Calendar invites, which were then triggered by the Gemini AI assistant. The AI triggered pre-programmed actions, including controlling smart home devices without the users’ asking.
- Malicious instructions were embedded within a seemingly innocent prompt or object.
- The indirect prompt injection technique was used to inject malicious code.
How this affects you
This was a controlled experiment to demonstrate the vulnerability in Gemini. It was not an actual live hack. However, it highlights the potential risks and the need for Google to take action to prevent such attacks.
“Even if the impact was real, it was done as a controlled experiment to demonstrate a vulnerability in Gemini.” – A researcher involved in the project
What you can do to protect your devices
While this attack was specific to Gemini and Google Home, the following general recommendations can help protect you and your devices from cyberattacks.
| Limit permissions | Don’t give Gemini, Siri, or other smart home assistants control of sensitive devices unless you need to. |
| Be mindful of services | The more devices and apps you connect to your AI assistant, the more potential entry points there are for attackers. |
| Watch for unexpected behavior | If something seems off, revoke permissions and report it. |
Additional tips
- Keep your devices and apps up-to-date with the latest firmware updates.
- Use antivirus software to protect against malware.
Conclusion
This demonstration highlights the need for Google to prioritize the security of its AI models and devices. While it’s unlikely that someone will try to hack your smart home devices, taking precautions and being aware of potential risks can help prevent such attacks. By following the recommended tips and staying informed, you can help protect your devices and your data.