MOUNTAIN VIEW, Calif. — Google has indefinitely postponed the rollout of Gemini AI-powered features for Android Auto after internal testing revealed potential driver distraction issues, the company announced Tuesday.
The features, first previewed at Google I/O in May, would have allowed drivers to use natural language to control navigation, messaging, and media playback. However, during simulated driving tests, the AI assistant reportedly misinterpreted commands 12% of the time — leading to driver frustration and extended eyes-off-road time.
"Safety is our north star," said Hiroshi Lockheimer, Google's senior vice president of platforms and ecosystems, in a blog post. "We will not release any feature that does not meet our rigorous distraction standards."
The decision comes as the National Highway Traffic Safety Administration (NHTSA) intensifies scrutiny of in-car AI systems. Last month, the agency opened a preliminary investigation into Tesla's voice command system after receiving 758 complaints about unresponsive or incorrect responses.
Independent testing by Consumer Reports found that drivers using voice assistants took an average of 22 seconds to complete simple tasks like changing a playlist or finding a restaurant — compared to 10 seconds using physical buttons.
"Voice assistants are supposed to reduce distraction, but poorly implemented ones do the opposite," said Jake Fisher, senior director of auto testing at Consumer Reports. "Google made the right call here."
Google did not provide a revised release timeline. The company confirmed that standard Android Auto voice commands — those not powered by Gemini — will continue functioning normally.
Competitors are moving faster. Amazon recently expanded its Alexa Auto integration to Ford and BMW models, while Apple's next-generation CarPlay, expected in late 2026, will feature deep Siri integration with vehicle controls.
Shares of Alphabet fell 1.2% following the announcement, while Google's AI safety team reportedly expanded by 40% this quarter.