Amazon, Apple, and Google all sent private voice recordings to thousands of human contractors for review — including intimate moments

When you speak to Alexa in your kitchen, who's really listening? For years, Amazon, Apple, and Google assured users that their voice assistants were private—that recordings existed only as data streams processed by algorithms. That promise crumbled in April 2019, when Bloomberg revealed what the companies had kept quiet: thousands of human beings were reviewing those recordings, many captured in the most intimate moments of users' lives.

The revelation came as a shock, but not because the practice was new. The companies had been employing contractors worldwide to listen to voice recordings for months, if not longer. What was shocking was that they hadn't told anyone. Users activated their devices assuming privacy protections were in place, unaware that confidential medical information, drug deals, and recordings of couples having sex were being heard by strangers in offices thousands of miles away.

When confronted, the companies offered a familiar defense: this was just quality assurance. Amazon, Apple, and Google all claimed that human review was necessary to improve their AI systems—a reasonable-sounding explanation that obscured a fundamental truth. Users had never consented to having their intimate moments audited by contractors. The companies had simply decided that the technical benefits outweighed transparency.

But the details made the justification harder to swallow. Amazon's practice wasn't limited to recordings where Alexa was deliberately activated. Workers heard unintentional activations—devices triggered by words that sounded like "Alexa" or by accident when someone reached for the device. Apple contractors listened to roughly 1,000 Siri recordings daily, most of them never explicitly requested by users. Google Assistant operated under similar conditions. The scale was staggering, the consent was nonexistent, and the potential for abuse was obvious.

Evidence of the companies' initial indifference to disclosure is telling. They didn't volunteer this information. They didn't update their privacy policies to mention it. They didn't alert users that humans were hearing their conversations. Only when journalists began asking questions did they acknowledge the practice at all. Even then, their responses were minimal, treated as technical details rather than fundamental privacy breaches.