Meta has disabled camera access on thousands of AI glasses after detecting capture-light tampering. The white LED identifies photography, video and livestreaming, although some camera-based AI functions can operate without it.
Updated 2 September 2026: This article now includes Meta’s latest enforcement figures, its fix for recordings started before the LED was covered and its public-awareness campaign.
Meta has disabled cameras on thousands of its AI glasses after detecting attempts to cover, modify or destroy the white capture LED.
Alex Himel, Meta’s VP of Wearables, told Semafor that fewer than 0.1% of all glasses sold had been tampered with. Meta hasn’t disclosed the precise number of affected devices or identified which models were involved.
The response depends on the type of interference. Business Insider reports that a camera blocked by a sticker can work again once the obstruction is removed. Physically destroying or drilling out the LED can permanently disable the camera.
Meta has also closed a separate workaround. Its previous safeguard checked whether the LED was covered when recording began, allowing someone to start filming before obstructing the light. An update announced on 27 August stops an active recording if the LED is subsequently covered.
The enforcement deals with deliberate interference. It doesn’t change what the light represents during normal use.
Meta told The XR Beat:
“A capture LED will blink bright white when you take a photo or video you can save or share.”
Ray-Ban’s German guidance says the LED also signals livestreaming. Camera-based AI functions can operate without the light on newer models because their images aren’t intended for the wearer’s gallery. First-generation glasses may still illuminate the LED during some AI functions because of hardware limitations.
One Camera, Several Functions
Ray-Ban Meta glasses combine photographs, video, visual AI, video calls and, on some models, livestreaming. The XR Beat has separated camera glasses, visual AI and display eyewear into distinct product categories because their data flows differ.
Meta’s July 2026 privacy explainer says the capture light blinks for a photograph and remains on during video. Meta says it has no off switch and that, beginning with its second generation of glasses, the camera is disabled when a blocked capture LED is detected. On 27 August, Meta said an additional update had started rolling out to stop an active recording if the light is covered after capture begins.
Meta gave The XR Beat the same description in an attributable statement:
“A capture LED will blink bright white when you take a photo or video you can save or share.”
Ray-Ban’s current FAQ says camera-based AI on newer models can operate without the light when ‘using camera-based AI functions’, and give the example of identifying a landmark or plant. First-generation glasses may still illuminate it during some AI functions because of hardware limitations.
Visual AI still uses the camera. Meta’s multimodal AI system card says a visual request can trigger a photograph, with the image and wearer’s words sent for model processing. Last updated in December 2023, it says the capture light indicates camera use. It predates Gen 2 and doesn’t cover every model, software version or AI function.
Meta didn’t confirm on the record whether the light stays on during camera-based AI requests on Wayfarer Gen 2 or reconcile its position with Ray-Ban’s FAQ. Ray-Ban owner EssilorLuxottica hadn’t responded.
Hamburg Checked Wayfarer Gen 1 and Part of Gen 2
The Hamburg Commissioner for Data Protection and Freedom of Information began examining the glasses in summer 2025. Its activity report covers capture visibility, AI, microphones, data transmission and possible use of data for AI training.
The authority bought two pairs, checked the light and data flows, and opened one pair for component analysis. A separate technical and legal report is due in the coming weeks.
Spokesperson Eva Zimmermann told The XR Beat that the office tested Ray-Ban Meta Wayfarer Gen 1 and “partially” tested Gen 2. Its latest checks used Android and Meta AI app version 22.9.0.0.29, including camera-based Meta AI. Zimmermann said its conclusion covered photograph and video recording, adding that:
For camera based AI there is no LED signal at all.
Hamburg didn’t supply the glasses’ firmware version, lighting conditions, viewing angles, a feature-by-feature record or a definition of its partial Gen 2 work. It gave no separate results for video calls or livestreams, so its findings can’t describe each model or configuration.
Ray-Ban’s guidance and Hamburg’s checks align on the observed behaviour. Hamburg said it’s in contact with the Bundesnetzagentur and Ireland’s Data Protection Commission. Its forthcoming report should provide the method, fuller results and legal analysis.
German Law Hasn’t Classified Visual AI
Section 8 of Germany’s Telecommunications Digital Services Data Protection Act, known as the TDDDG, covers networked devices disguised as everyday objects when they’re especially suited and intended to capture another person’s image or non-public speech without the person noticing. Intended use can count as unnoticed capture when the function isn’t clearly recognisable to the affected person.
The Bundesnetzagentur tells consumers that smart glasses need an optical or acoustic warning when photographing or filming. Its guidance doesn’t explain whether an image transmitted for an AI response falls into the same category when it isn’t saved to the gallery.
Meta said its glasses passed a Bundesnetzagentur review in 2022. The agency confirmed that earlier assessments covered various Ray-Ban and Oakley smart glasses developed with Meta. Wayfarer Gen 2 wasn’t included.
The agency can’t currently assess Meta’s AI functions under section 8. It hasn’t opened a formal smart-glasses proceeding and didn’t identify the models, dates or results from its earlier work.
HateAid filed a criminal complaint in August alleging a TDDDG breach. Frankfurt’s Cybercrime Center, known as ZIT, is conducting a preliminary examination to decide whether sufficient initial suspicion exists. No formal investigation has opened, and no authority or court has declared the glasses illegal or ordered a German sales ban.
Employers Will Need Function-Level Rules
Camera-based AI can support translation, object identification, scene descriptions and hands-free assistance. It can help disabled users and workers who need information without reaching for a phone.
Meta’s smart glasses sales are growing, increasing the chance that consumer devices will enter offices, factories, hospitals, shops and customer sites. Employers need to decide which models and functions suit each environment before approving them.
A policy based only on the white light won’t cover each camera use. Buyers should document where gallery capture, visual AI, video calls and livestreams are permitted; how workers and visitors are informed; which images and prompts leave the device; where data is processed; how long it’s retained; and whether supplier terms allow workplace data to improve products or models.
Controls on the companion phone may cover application installation and network access while leaving glasses firmware, personal Meta accounts and cloud AI settings outside the employer’s reach. Pilot teams should retest those boundaries after firmware and app updates.
Bathrooms, changing rooms and other sensitive spaces need separate rules because a camera-equipped device can create risk regardless of the active function.
Hamburg’s final report may clarify its method and how German law treats visual AI. Meta’s latest update closes the workaround of starting a gallery recording before covering the light. Camera-based AI on newer models can still run without the white light, so employers need written controls for each camera function.
