Many people learned the hard way that “Hey Google” doesn’t always stay silent. Google has agreed to a $68 million settlement after users said its voice assistant captured private conversations without clear consent.
The case shows how a convenience feature quietly became a privacy risk for millions of device owners. It also highlights how often everyday speech can be misheard by machines. And it puts a spotlight on what companies do with audio once it is collected.
The dispute centers on Google Assistant, the AI helper built into many Android phones, smart speakers, and connected devices. Assistant is designed to sit in standby mode until it hears a trigger phrase such as “Hey Google.”
Once activated, it records audio and sends it to Google’s servers for analysis. Plaintiffs argued that the system frequently misheard ordinary speech as a command, a problem known as “false accepts,” and began recording when no one intended to activate it.
According to court filings, some unintended recordings involved sensitive talks about finances, personal decisions, and workplace matters. Users further alleged that parts of this audio were later used to refine targeted advertising. Google has denied secretly recording or selling conversations, but acknowledged that mistaken activations can occur.
Meanwhile, the proposed settlement was filed in federal court in San Jose and still requires approval from U.S. District Judge Beth Labson Freeman. If approved, it will operate as a class action covering people who owned Google devices or experienced false accepts since May 18, 2016.
Google would place the full $68 million into a compensation fund that covers consumer payouts, administrative costs, and court-approved legal fees. Plaintiffs’ lawyers may seek up to one-third of the total, roughly $22 million.
Furthermore, this agreement follows a similar legal battle involving Apple. In January, Apple agreed to pay $95 million to settle claims that Siri recorded private conversations without permission.
