Ray-Ban Meta Glasses Were Recording Audio Without the 'Hey Meta' Command

On October 16, 2026, Ars Technica reported that a software bug in Ray-Ban Meta smart glasses caused the device to record audio without its owner's consent. According to complaints that appeared on Reddit three days earlier, the glasses activated Meta AI on their own and recorded nearby conversations even though the 'Hey Meta' voice command had not been used — in some cases the feature was not enabled at all. Users found the unexpected recordings inside the Meta AI mobile app — as saved audio files and their text transcripts.

The report describes a device malfunction: the glasses began recording audio without the wearer's consent. The complaints were also covered by 9to5Google (October 15) and TechSpot (October 16): both outlets confirmed the incident by compiling the Reddit posts.

How the Complaints Started

The first complaints were posted on Reddit on October 13, 2026. According to users, the glasses woke Meta AI and started recording audio without any voice command. Normally, Meta AI should only activate when the phrase 'Hey Meta' is spoken — the glasses' primary voice control mechanism.

One complainant reported observing this behavior on three separate pairs of glasses, including second- and third-generation models. This detail shows the problem was not limited to a single defective unit: the same behavior repeated across multiple devices and generations.

Another important detail: some users had the 'Hey Meta' response feature turned off entirely. In other words, the glasses theoretically should not have responded to a voice command at all, yet they still woke Meta AI and began recording. This suggests the malfunction was independent of individual settings. Its presence on both second- and third-generation models indicates a software-level bug regardless of the hardware generation.

Where the Recordings Were Found

Users discovered the unexpected recordings inside the Meta AI mobile app. The app displayed both the saved audio recordings and their automatic text transcripts. So the glasses not only recorded audio — the recordings were transmitted to the Meta AI system, transcribed into text, and stored in the user's account.

While recording, the white indicator light on the glasses' inner frame was lit. The light is a privacy feature meant to warn bystanders that recording is in progress. Even so, the owners themselves did not notice when recording began — the glasses started recording mid-conversation, with no command at all. A lit indicator did not mean the recording had started with consent.

Timeline: What Happened in Three Days

Events unfolded over three days. On October 13, the first complaints appeared on Reddit, with users reporting that their glasses were recording without a command. On October 15, 9to5Google published an article compiling the complaints, and on the same day a Meta spokesperson confirmed to Android Authority that the problem was 'resolved.' On October 16, Ars Technica covered the bug more broadly, and TechSpot confirmed the incident through its own sources.

This speed shows the issue moved from user complaints to professional media attention within two days. Meta's official comment came two days after the complaints appeared — on October 15.

Meta's Response

A Meta spokesperson confirmed to Android Authority that the issue had been fixed:

The issue is resolved. Users can view, manage, and delete audio recordings of their interactions with Meta AI at any time through settings.

The company disclosed neither the precise technical cause of the bug nor the number of users affected. Meta's statement was limited to confirming the fix and the ability to review and delete audio recordings. The company acknowledged the malfunction but provided no information about its scale or root cause. The statement said nothing about when the bug appeared, what changes were made to fix it, or whether affected users were notified.

What Remains Unknown

According to Android Authority, the number of complaints remained limited, and the bug was not independently reproduced by third-party sources. The issue was observed to have appeared after the latest software update, though there is no evidence that the update was the direct cause. The limited number of complaints suggests the problem did not affect all users, but the exact scope remains unknown.

So three questions remain open for now: what was the technical cause of the malfunction, how many users did it affect, and what exactly does the 'resolved' statement mean — which specific fix it refers to. Meta has not yet answered these questions.