A Bug Let Some Ray-Ban Meta Glasses Record Audio Without the "Hey Meta" Wake Phrase
Meta says the issue is fixed, but it hasn't explained what caused it or how many wearers were affected.
Some Ray-Ban Meta smart glasses were listening even when their owners never said a word to them. This week, users reported that the glasses recorded audio without anyone triggering them with the "Hey, Meta" wake phrase, which is supposed to be the only way to activate the AI assistant's microphone for a conversation.
One Reddit user, posting as Kindly-Audience-137, said the problem happened across three separate pairs of Meta Ray-Bans, despite having turned off the "Respond without Hey Meta" setting that lets the glasses listen for follow-up questions without the wake phrase. According to the user, the Meta AI app was saving both audio recordings and text transcripts of these unprompted interactions directly into their conversation history.
"I believe this deserves a serious technical investigation," the user wrote. "Customers should not have to wonder whether their private conversations are being captured when they have not intentionally activated their glasses' AI assistant."
Android Authority reported on the discussions Tuesday and later published Meta's response: "This issue has been resolved. People can manage, review, and delete voice logs of their interactions with Meta AI via settings at any time."
That statement confirms something went wrong, but it doesn't explain what. Meta hasn't said what caused the glasses to start recording without a wake phrase in the first place. Kindly-Audience-137 said the behavior appeared after a recent firmware update, but Meta hasn't confirmed or denied that a bad update was the trigger. The company also hasn't disclosed how many devices were affected. Android Authority said it couldn't reproduce the bug itself before Meta announced the fix, leaving the scope of the problem murky — a handful of Reddit reports, or something wider that simply hasn't surfaced yet. Ars Technica asked Meta directly how many devices were hit and what it's doing to prevent a repeat; as of publication, Meta hadn't responded.
Why this bug lands differently
Software bugs happen. What makes this one sting is that it struck at the exact fear privacy advocates have been raising since Meta started selling camera- and microphone-equipped glasses that look like ordinary Ray-Bans. The whole pitch of the product is that it doesn't look like surveillance equipment. There's no obvious lens housing, no bulky headset, nothing that signals "this thing might be recording you." That's a selling point for wearers who want hands-free photos and AI help without looking like they're wearing a gadget. It's also exactly why critics have warned the glasses make it unusually easy to record someone without their knowledge.
Meta has built in some signals meant to address that: an LED indicator lights up when the camera or microphone is actively capturing. But a bug that activates recording without the wake phrase undercuts the entire premise that the system only listens when the wearer deliberately invokes it. If a firmware glitch can quietly flip the glasses into listening mode, the assurance that the hardware only engages on command is only as solid as the code behind it.
The trust problem Meta can't fix with a patch
Even a small number of affected units is enough to raise the stakes, because the product being sold here depends almost entirely on a promise: the glasses don't record unless you tell them to. That promise doesn't just protect the wearer. It's the thing that's supposed to reassure everyone else — the people sitting across the table, walking past on the street, standing next to someone wearing these glasses — that they aren't being secretly captured and uploaded to an AI system's conversation history.
Meta's one-line statement says the issue "has been resolved," which is reassuring as far as it goes, but it leaves the harder questions open. What exactly broke. Whether it's tied to the firmware update the Reddit user flagged. How many people's conversations got recorded and stored before anyone noticed. And whether Meta has changed anything structurally to prevent the next version of this bug from happening with the next firmware push.
Meta does offer tools to review and delete voice logs after the fact, which matters for the people who know to check. It does nothing for the people who were recorded by someone else's glasses and have no idea a conversation they were part of ended up as a transcript in a stranger's app. That's the part a software fix can patch over but can't fully resolve. Smart glasses ask people to trust a company's code to only listen when invited. This week, for at least some users, that code didn't hold up, and nobody outside Meta yet knows exactly why.
