Meta Smart Glasses Recorded Audio Without the Wake Phrase — Cybersecurity Alert

What are Meta's smart glasses?

Meta's smart glasses (made with Ray-Ban) are camera-and-microphone glasses with a built-in AI assistant. They are supposed to record audio only after hearing the "Hey Meta" wake phrase — but in October 2026, owners found recordings and transcripts of conversations nobody asked the assistant to capture.

What happened

• ~Oct 13, 2026 — Reddit users reported finding ordinary, everyday conversations saved in the Meta AI app: audio recordings plus text transcripts — on glasses where the assistant had never been intentionally activated.
• Oct 13, 2026 — One owner's report (per Techtimes UK): the "Respond without Hey Meta" setting was disabled, the behavior spanned three separate pairs (2nd- and 3rd-generation Ray-Ban Meta glasses), and the glasses' LED notification light appeared to activate during normal conversation. The user first noticed it after a recent firmware update — there is no evidence the update caused it.
• Oct 13, 2026 — Meta told Android Authority the issue "has been resolved". It named no cause, no firmware version, and no count of affected owners.
• Oct 16, 2026 — PCMag surfaced the story: Meta admits its smart glasses briefly recorded audio without a wake phrase, and has not explained what caused the issue.

Honest scope: Android Authority reported the complaints remained limited, were not independently reproduced, and there is no evidence the problem was widespread. Treat this as a real but limited incident — not a confirmed mass surveillance event.

What was recorded

• Audio recordings and written transcripts of conversations the wearers say they never asked Meta AI to capture.
• They appeared in the Meta AI app's activity history — visible to the glasses' owner, but recorded from conversations with people around them who never consented.
• Security researchers at IPVM examined related cases where the LED indicator either failed to light consistently or recording began before the wearer finished the wake phrase — meaning audio could be captured before the visible warning caught up.

The US vs UK privacy notice difference

Reporting on Meta's AI Glasses Voice Privacy Notice found the US and UK versions differ:

• US: Meta will "store voice recordings even if you unintentionally activate a voice interaction."
• UK: with storage off, recordings are deleted "immediately after processing."

Check the version that applies in your region — the protections are not the same everywhere.

The bystander problem

The LED indicator is the entire social contract that makes camera-and-microphone glasses tolerable around other people: you can at least see when you are being recorded. This incident broke that contract twice over — recordings happened without the wake phrase, and the indicator reportedly didn't light reliably. For the people around a wearer, who never agreed to be recorded, that's the serious part.

Background: the glasses were already restricted in some New York and UK courtrooms over filming-without-consent concerns, and Norway is considering stricter measures on their use in public spaces.

What to do now

1. Check your Meta AI activity log for recordings you didn't trigger. If anything looks unintended, note the date.
2. Review and delete voice logs in the Meta AI app settings — audio and transcripts.
3. Turn OFF "Respond without Hey Meta" if it's on — this feature lets the assistant stay active on natural conversation without the full wake phrase.
4. Keep firmware updated — Meta deployed the fix through a software update.
5. Tell people around you what the LED indicator means — and that it hasn't always worked reliably.

Are your smart glasses behaving?

Probably not as well as you assume. Meta confirmed its glasses recorded audio without the wake phrase, and the LED indicator did not always warn bystanders in time. Answer these 5 questions to find out whether your pair respects your privacy — or records first and asks later.

Cybersecurity FAQ