Loading story…
Meta says it has resolved an issue following reports that Ray-Ban Meta smart glasses recorded and transcribed conversations without users saying “Hey Meta” or knowing recording had begun. One owner reported unexpected recordings across three pairs of glasses after a firmware update, despite having “Respond without Hey Meta” disabled. Other commenters described similar unexplained activations, although the reports do not establish how widespread the problem was.
The behavior raises privacy concerns for wearers and bystanders, particularly where glasses are used around confidential workplace conversations. Meta has not publicly explained the cause, the number of affected users, or what happened to the captured audio. Users can review and delete voice-interaction logs through settings; recommended precautions include checking voice history for unexpected recordings, reviewing activation settings, and keeping the glasses and companion app updated.
See affected versions and whether adversaries are exploiting it.
3 events from the most recent confirmed update back to the earliest known activity.
Meta told Android Authority that the issue had been resolved and said users could review and delete Meta AI voice logs through settings. The company did not publicly explain the cause, the number of affected users, or what happened to the captured audio.
According to the Reddit owner, Meta Support recommended disabling microphone and camera permissions, turning off Wear Detection, deleting voice history, and disabling automatic updates. These restrictions would limit much of the glasses’ core functionality.
Owners reported discovering recordings and transcripts of conversations captured without saying “Hey Meta.” One Reddit user reported the behavior across three pairs after a firmware update, including with “Respond without Hey Meta” disabled; the update was not established as the cause.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
See whether adversaries are exploiting this yet, and where the affected versions run in your environment.
3 references tracked. Mallory keeps watching after this page renders.
malware.news
Open sourcemalwarebytes.com
Open sourcereddit.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Investigations reported by Swedish outlets *Svenska Dagbladet* and *Göteborgs-Posten* found that recordings captured by **Meta Ray-Ban smart glasses**—including video and audio—are being reviewed by human contractors as part of AI training and quality assurance workflows. Workers employed by **Sama**, a Meta subcontractor in **Nairobi, Kenya**, described routinely handling highly sensitive content inadvertently recorded by users, including bathroom visits, undressing, sex/pornography, and private conversations, as well as incidental capture of **bank cards** and other identifying details; interviewees said they feared reprisals for raising concerns and described strict on-site controls intended to prevent leaks. Following the reporting, the UK’s privacy regulator, the **Information Commissioner’s Office (ICO)**, confirmed it is contacting Meta to ask questions about the devices and associated data-handling practices. While Meta’s terms reportedly disclose that some interactions may be reviewed by humans to improve the system, the reporting and worker accounts suggest the review pipeline can include intimate or identifying moments that wearers may not expect to be viewed by third parties, raising regulatory and reputational risk around consent, transparency, and safeguards for bystander and user privacy.
Jun 29, 2026
Meta's Ray-Ban smart glasses are facing growing restrictions from pubs, restaurants, and theatres as venue operators respond to privacy concerns over devices that can discreetly record audio and video while resembling ordinary eyewear. UK organizations including **Soho House**, **ATG Theatres**, and **Wetherspoons** have reportedly banned or limited the glasses, citing risks to patrons and staff, while reports also said footage captured through Meta's AI features was sent to Meta servers and reviewed by human contractors at **Sama** in Nairobi for AI training, including highly sensitive scenes such as people undressing and visible bank cards. The backlash has intensified after a newly published Meta patent application described smart glasses that could pair AI with facial recognition to identify nearby people and automatically generate media content about them. Although the filing does not confirm the feature will be released, privacy advocates warned it could enable stalking, harassment, and abuse, especially against vulnerable people, and renewed scrutiny of Meta's earlier work on facial-identification features such as the internally developed **"Name Tag"** capability and previously disabled facial-recognition code in Meta AI.
Aug 18, 2026
Leaked documents and app analysis indicate Meta was developing a facial-recognition feature called **NameTag** for its Ray-Ban smart glasses and the Meta AI companion app. Reporting cited in the references says the software included on-device components for face detection, face cropping, liveness checks, and biometric template generation, with the apparent goal of identifying people seen through the glasses by linking them to Meta platform data such as Facebook and Instagram profiles. The code was reportedly distributed inside the Meta AI app to millions of smartphones even though the feature was not publicly launched. After scrutiny and backlash, Meta reportedly removed most of the related code from a newer version of the app, but privacy concerns persisted because the capability could be restored and because bystanders would have no meaningful way to opt out of being scanned. The reports also say Meta licensed facial-recognition and liveness-detection technology from **Rank One Computing**, a vendor known for government and law-enforcement work, and that the software could support databases of up to **10 million** biometric profiles, intensifying concerns about consumer surveillance through wearable devices.
Jul 15, 2026