Senators are looking into 'serious privacy risks' with Meta's AI glasses

Meta's AI smart glasses are wearable devices fitted with cameras and microphones. They can capture imagery and conversations while a person wears them. Their AI features can process or respond to captured imagery, making the glasses more than ordinary eyewear.

The core mechanism is simple but powerful: cameras collect visual information, microphones collect sound, and AI interactions may use that material. The article says Meta may blur faces and license plates for some AI interactions, as it does on Facebook and Instagram. It also asks when images become accessible to employees or contractors.

The article does not provide a complete technical specification or list every AI function. It does show why the design matters. Glasses can record people nearby, including bystanders who did not choose to participate. Senators are now asking Meta about accidental activation, recording-indicator tampering, privacy testing, and whether captured imagery trains AI models.

Wyden and Markey are investigating how Meta's AI glasses may threaten users' and bystanders' privacy and safety. They want Meta to explain how the glasses record, protect, store, and use captured material. Their inquiry reaches beyond individual misuse to Meta's own product choices and oversight.

The senators cited reports of men secretly filming women, harassing them, and posting recordings online. They also cited reports and a lawsuit alleging that AI contractors reviewed footage showing people undressing and engaging in sexual activity. Their letter asks about accidental activation, recording-indicator tampering, face and license-plate blurring, employee access, and AI training.

They are questioning Meta because the company reportedly identified bystander-privacy risks before launching the product. They also ask whether Meta performed detailed human-rights due diligence and safety testing. Meta was asked to respond by November 26, 2026. The senators say privacy commitments must produce meaningful safeguards.

The article identifies five groups connected to the glasses' privacy risks. Wearers may have conversations recorded. Bystanders may be filmed without knowing it. Meta employees and contractors may gain access to captured imagery. People who see recordings online may encounter material involving others without their consent.

The pathway is what makes the risk broad. A wearer or another person can activate the camera, intentionally or accidentally. The resulting imagery can then be reviewed by company personnel, used in AI interactions or training, or posted to social media. Each step creates another group with potential exposure to the material.

No headcount or number of affected people is given. The article instead describes several routes of access and harm. It cites secret filming, intimate footage reviewed by contractors, and recordings shared online. The senators' questions focus on controlling these routes before private material travels further.

Several incidents drove concern about Meta's glasses. The senators cited 404 Media and BBC reports about men secretly filming women with the devices, calling them “pervert glasses,” while harassing the women. The men then posted recordings to social media.

The article also mentions reports and a lawsuit involving Meta's AI contractors. According to those accounts, contractors said they reviewed footage captured by the eyewear that showed people undressing and engaging in sexual activity. This raised questions about how intimate material reached reviewers and whether access controls worked.

Other complaints involve the glasses recording conversations without consent after a firmware update. The senators separately asked Meta about accidental camera activation and tampering with recording indicators. These reports create different pathways to hidden recording: deliberate misuse, possible software-related recording, and weakened signals that might otherwise alert people nearby.

Recording someone without consent can remove their control over personal images, words, and intimate moments. The article links the glasses to disturbing invasions of privacy, harassment, and recordings posted to social media. People may therefore face unwanted public exposure after believing they were in a private or ordinary situation.

The examples are especially serious because some footage reportedly showed people undressing or engaging in sexual activity. According to reports and a lawsuit, AI contractors said they reviewed such material. Conversations may also be captured without consent, including after a firmware update. These pathways can expose sensitive information to strangers, company personnel, or online audiences.

The article does not list every possible legal or personal consequence. It does show that harm can continue after recording. Images may be shared, reviewed, or used to train AI models. The senators want Meta to explain these practices and demonstrate meaningful steps to prevent its products from facilitating intimate-image abuse.

Technical safeguards should make recording visible and reduce unnecessary identification. Clear recording indicators could alert nearby people, while protections against tampering could preserve that warning. Face and license-plate blurring could limit exposure during some AI interactions. These measures would not replace consent, but they could reduce accidental or unnecessary capture.

Company safeguards should tightly limit who can view imagery. Meta could explain when employees or contractors receive access, keep intimate material out of routine review, and separate captured images from AI training unless appropriate permission exists. The article shows why these controls matter: contractors reportedly reviewed footage involving nudity and sexual activity.

Legal and workplace rules should prohibit secret recording in private situations and harassment using the devices. The article does not describe a complete legal framework. It does show that safeguards need testing before launch. Meta should address accidental activation, indicator tampering, consent, access, blurring, and human-rights due diligence.

Consent gives people a say over whether their image, voice, or intimate activity is captured. Control also includes knowing that recording is happening and understanding who may access the material. Without those choices, people can lose privacy even when they never bought or used the glasses.

The article illustrates the problem through secret filming, recordings shared online, contractor access, and possible AI training. Bystanders may not know when a camera is active. They may also have no practical way to stop a recording or retrieve material after it spreads. Clear signals, limited access, and careful data use can help preserve control.

Human-rights due diligence and safety testing should happen before release because product risks can affect people outside the customer relationship. The senators cite Meta documents and ask whether detailed checks occurred. Their question matters now because Meta reportedly knew about bystander-privacy risks before launching the glasses.