CYBERSECURITY RESEARCHERS WARN OF FLAW IN GOOGLE'S GEMINI CHATBOT THAT COULD LET HACKERS SEIZE SMART HOME DEVICES

A team of cybersecurity researchers has uncovered a serious flaw in Google's Gemini AI chatbot that could potentially allow hackers to take control of connected smart home devices.

- 

The vulnerability, detailed in a new report shared with regulators and disclosed to Google, raises urgent questions about the safety of AI systems that are increasingly embedded in daily life.

THE VULNERABILITY

According to the researchers, the flaw centers on how Gemini interacts with third-party smart home platforms through natural language commands. Under certain conditions, malicious prompts can be crafted to trick Gemini into executing unauthorized actions. In practical terms, this means an attacker could potentially manipulate the chatbot into unlocking doors, disabling alarms, or controlling appliances linked to a user's Google Home ecosystem.

Unlike traditional hacks that require breaking through multiple layers of network security, this exploit relies on the AI's tendency to interpret and execute complex instructions without sufficient verification safeguards. In tests, researchers demonstrated scenarios in which Gemini could be manipulated into granting access or bypassing restrictions by responding to carefully engineered requests.

POTENTIAL IMPACT ON SMART HOMES