Don’t use Gemini Live while driving, warns Google founder Sergey Brin

google

Google co-founder Sergey Brin has issued a clear warning to users: don’t rely on Gemini Live while driving, at least not yet. Speaking at a recent panel discussion at Stanford University, Brin said the current public version of Gemini Live is based on an outdated model and is not ready for use in situations that demand full attention, such as driving.

According to a report by Business Insider, Brin admitted that he personally tests Gemini Live while commuting. He described this as “dogfooding,” a term used in the tech world to mean using your own product to understand how it works in real-life situations. During these tests, Brin said he asks the AI complex questions, including estimating the cost and power needs of building data centres.

However, Brin made it clear that his personal use should not be taken as advice for others. “You shouldn’t do it now, because we have a way better version coming,” he said, explaining that the model currently available to users is what he called an “ancient” one.

Brin stressed that there is a big gap between the version of Gemini Live available to the public and the one Google is testing internally. He said the internal version is far more capable and handles complex questions much better. A significantly improved public release is expected within weeks and will likely be powered by Gemini 3, Google’s next-generation AI model.

The warning comes at a time when more people are using voice-based AI tools in everyday situations, including while commuting. Brin’s comments suggest that Google is aware of the risks of distraction and wants users to wait until the technology is more reliable and natural before using it in high-risk scenarios.

Gemini Live itself is designed as a conversational AI experience, similar to Google Assistant but more advanced. It allows users to speak naturally, interrupt responses, and ask follow-up questions without starting over. Users can also choose from multiple voice options and adjust the tone of responses.

For example, someone preparing for a job interview can use Gemini Live to practise mock interviews or get tips on which skills to highlight. The idea is to make conversations with AI feel less robotic and more human.

For now, though, Brin’s message is simple: Gemini Live is improving fast, but users should wait for the upcoming update before trusting it in situations like driving, where focus and safety matter most.