Meta says an issue that led some Ray-Ban Meta owners to find ORDINARY CONVERSATIONS saved in their Meta AI history, WITHOUT ever saying “Hey Meta,” has been RESOLVED. The company gave that ONE-LINE assurance to Android Authority on Tuesday, October 13, roughly THREE AND A HALF HOURS after the outlet published the owner reports. What Meta did NOT say is what CAUSED the behavior, which FIRMWARE was involved, or what happens to audio that was captured BY MISTAKE.
What Owners Reported
According to Android Authority’s original report (10:49 AM ET, October 13), one Ray-Ban Meta owner said in TWO Reddit threads that THREE separate pairs, spanning SECOND- and THIRD-generation models, captured everyday conversations with NO wake phrase. Those exchanges reportedly appeared in the Meta AI app as SAVED audio recordings and TEXT transcripts, even though the “Respond without Hey Meta” setting was DISABLED. The owner also said the glasses’ internal NOTIFICATION LIGHT sometimes turned on during normal conversation, and that the behavior STARTED after a recent firmware update.
SEVERAL commenters described similar UNEXPLAINED activations. Android Authority was CAREFUL on scope: it said reports “remain limited,” that it had NOT independently reproduced the behavior, and that there was “no evidence to suggest this is widespread.”
Big surprise: Some Ray-Ban Meta owners say their glasses are listening without being asked https://t.co/q7DNrfhgqt
— Android Authority (@AndroidAuth) October 13, 2026
Notebookcheck, publishing at 2:08 PM ET, took a STRONGER line. It wrote that “judging by the sheer number of user reports, this issue is widespread,” noted that it affects BOTH Gen 2 and Gen 3 hardware, and said a HARDWARE DEFECT in only a few pairs was therefore UNLIKELY. It also reported that when one user asked, Meta SUPPORT said disabling the glasses’ access to the microphones and cameras would solve the problem, a workaround that, as the outlet put it, would leave the glasses with “virtually no functionality.” The two outlets DISAGREE on scale; NEITHER cites a count of affected devices.
Meta’s Answer, in Full
Android Authority updated its story at 2:17 PM ET with Meta’s statement:
“This issue has been resolved. People can manage, review and delete voice logs of their interactions with Meta AI via settings at any time.”
Meta, via Android Authority
That is the ENTIRE public explanation so far. Android Authority had asked whether Meta knew about the reports, whether a recent update was responsible, and what happens to audio captured through an unintended activation. The statement answers NONE of those questions directly; it CONFIRMS there WAS an issue and points owners to their LOGS.
What Meta’s Own Policy Says About Accidental Recordings
Meta’s AI Glasses Voice Privacy Notice (effective July 22, 2025) already anticipates MISACTIVATIONS. It states that Meta “will store voice recordings even if you unintentionally activate a voice interaction.” When its systems detect an unintended activation, those interactions are labeled “false wakes” and DELETED within 90 DAYS of detection. Otherwise, transcripts and stored audio are kept for up to ONE YEAR “to help improve Meta’s products,” and the notice says machine learning and TRAINED REVIEWERS process that data, with voice pitch altered during human review.
TWO details matter here. First, storage is ON BY DEFAULT for Ray-Ban Meta and Oakley Meta glasses, per the notice. Second, the policy covers “background sound that happens when you are interacting with our voice services,” so a conversation picked up IN ERROR is, by Meta’s own definition, voice data it may RETAIN until it is CLASSIFIED or deleted. Whether the recordings owners found were flagged as false wakes is something Meta has NOT addressed.
Meta’s help page for “Respond without Hey Meta” explains why that toggle is central to the reports: the feature is ON by default, and when enabled the MICROPHONE stays on after each request before turning off “after a little while.” The owners’ complaint is PRECISELY that capture CONTINUED with that feature switched OFF.
What Owners Can Do Now
Meta’s voice activity log guide gives the path: in the Meta AI app, tap Glasses, then Device settings, select the glasses, then Glasses privacy, then Voice activity log. Owners can REVIEW and DELETE entries there, and Meta’s Download Your Information tool EXPORTS stored transcripts, audio and related data such as the glasses’ serial number. Anyone who believes they were AFFECTED should CHECK the log for entries they did NOT initiate.
Why the Timing Is Awkward
The reports landed on the SAME DAY Meta put Ray-Ban Meta Audio on sale at $349, its first CAMERA-FREE glasses (our launch coverage). Dropping the camera ANSWERS the most visible smart-glasses privacy objection. It does NOT touch the MICROPHONE question, and a microphone-first product makes WAKE-WORD RELIABILITY more important, not less. The episode also arrives while REGULATORS are already circling the category, from Norway’s proposed AI-glasses restrictions onward.
Our read: Meta moved FAST, and a fix within hours is the RIGHT outcome. But “resolved” is a STATUS, not an EXPLANATION. Until Meta says what triggered the captures, which BUILDS were affected and whether the resulting recordings were treated as false wakes, owners have ONLY the log in the app to go on. Metaverse Watcher will UPDATE this story if Meta provides more detail.