Senators Press Meta on Smart Glasses Footage [2026]

On Wednesday, October 28, 2026, Senators Ron Wyden of Oregon and Edward Markey of Massachusetts sent Meta a letter demanding answers about the privacy risks of its Meta smart glasses, including what happens to bystander footage that may be used for AI training. 404 Media reported the letter on October 29, quoting the senators' warning that "Meta's smart glasses have increasingly been implicated in disturbing invasions of users' and bystanders' privacy." The letter draws together a year of reporting: harassment videos made with Ray-Ban Meta glasses, human reviewers who allegedly saw intimate footage, and a patent filing for facial recognition.

If you wear camera glasses, stand near someone who does, or publish video that includes strangers, this matters. The letter does not change the law, but it lays out in plain terms what the glasses can capture, where that footage may go, and which questions Meta has not yet answered publicly. This post walks through what the senators asked, what the underlying reporting says, what remains an allegation, and what you can do today, including how to blur the faces of bystanders before you post.

What Did the Senators Ask Meta in the Smart Glasses Letter?

The senators asked Meta how it prevents the nonconsensual capture, disclosure or processing of sensitive imagery, and what it does with footage from the glasses. According to 404 Media, the letter demands answers on four things in particular:

• What safeguards, if any, Meta has to prevent the nonconsensual "capture, disclosure, or processing of sensitive and intimate imagery."
• Whether Meta specifically evaluated the potential for its glasses "to facilitate voyeurism, nonconsensual intimate image capture, or similar privacy invasions."
• What imagery Meta uses for AI training.
• Whether people can request that their biometric and other data captured by the glasses be deleted, along with several more questions about the glasses' safety and design.

The letter also quotes Meta's own disclosures. In the senators' words:

"Meta's own disclosures confirm that imagery processed through its glasses' visual AI features may be retained and used to train AI systems, including with the assistance of human reviewers."

They go further and argue that training on this material is a privacy harm in itself:

"Using intimate imagery for AI development is itself an invasion of privacy when the people depicted have not consented to that use, as is inherently the case for most bystanders captured by Meta's glasses."

This is a letter, not legislation or a ruling. It asks questions and creates no deadline-backed obligation in the reporting we have seen. Its significance is that two senators with long records on surveillance and consumer privacy put the bystander problem on the congressional record in specific terms.

Who are Wyden and Markey, and is this their first letter?

No. Wyden and Markey, together with Senator Jeff Merkley, wrote to Meta in March 2026 about facial recognition plans for the glasses, as covered in our post on Meta's smart glasses facial recognition and the ACLU backlash. The new letter is a different topic: it concerns intimate imagery, human review and AI training, and it builds on a 404 Media investigation published in October 2025. 404 Media reports that investigation found Instagram accounts with hundreds of thousands of followers and millions of views posting videos filmed with Ray-Ban glasses of men entering massage parlors across the US and soliciting workers for sex.

What Has the Reporting Found About Smart Glasses and Bystanders?

The reporting describes three separate problems: harassment content made by wearers, human review of footage, and covert recording. Each has its own evidence trail, and it helps to keep them apart.

Harassment content filmed with the glasses

404 Media's 2025 investigation is the one the senators cite. It also reports that research published in August showed how Meta's platforms encourage an ecosystem of abusive content from pickup artists and influencers who use smart glasses to capture interactions with people minding their own business in public or at work, some of whom may not know they are being filmed. In a January report by the BBC, as summarized by 404 Media, a woman recorded with the glasses said she was "bombarded" with phone calls because her number was posted with the video, and men who had seen it stalked her at work. These are reported accounts; the details come from the outlets named.

The pattern matters for anyone who appears in public video: the harm usually comes not from the act of filming but from publishing, tagging and the details that make a person findable. Our guide to smart glasses misuse, bans and how to push back covers venue bans and practical responses.

Human reviewers and AI training

According to 404 Media, more than 70 people sued Meta earlier this year, claiming their intimate images and videos were exposed to workers hired by the company to label data for AI. In March, Swedish journalists reported that Kenyan data labelers were viewing and annotating uncensored footage from Meta's smart glasses, including sexual and violent content, and Meta ended its contract with those workers less than two months later, per the same report. Fortune covered the lawsuit in March. These are allegations in litigation and journalism, and 404 Media's report does not include a Meta response.

The senators' letter draws a line that is worth quoting, because it frames the actual legal and ethical question:

"Human review to fulfill a service knowingly requested by the person depicted is one thing; exposing unsuspecting individuals' intimate moments to strangers for AI training is quite another."

Covert recording and the indicator light

404 Media also notes that people have been modding these glasses to disable the light that indicates they are recording, for years. That is the detail most relevant to bystanders. A small LED is a weak form of consent: it is easy to miss in daylight, easy to cover, and, per that reporting, sometimes disabled outright. We looked at this tension in covert filming with smart glasses and how to protect yourself.

Why Are Bystanders at the Center of This?

Bystanders never agreed to anything, which is what separates the glasses from a phone held up in view. The wearer accepts Meta's terms. The person walking past, working a shift or sitting on a train does not. In the senators' phrasing, non-consent is "inherently the case for most bystanders captured by Meta's glasses."

Three things make the bystander problem harder than with phones:

1. Capture is ambient. The camera sits at eye level and can run for minutes without anything resembling the "pointing a phone at someone" signal people recognize.
2. Processing may leave the device. The senators cite Meta's disclosures that imagery handled by the visual AI features may be retained and used to train AI systems, with human reviewers in the loop. Footage that stays in a phone gallery has a different risk profile from footage that enters a training pipeline.
3. Identification is getting easier. In August, 404 Media reports, Meta filed a patent to add facial recognition and highlight reel features to its glasses. A patent is not a product, but it shows the direction. Our explainer on whether Meta blurs faces in smart glasses videos covers what the platform does and does not do today.

Regulators elsewhere have already reacted. A Dutch retailer paused sales, covered in Hans Anders halting Meta Ray-Ban sales in the Netherlands and Belgium, and Australia's eSafety regulator has pushed for stronger defaults, as in our post on eSafety and face blurring by default.

Update — November 7, 2026: Meta's product response is hardware segmentation: Ray-Ban Meta Audio drops the camera while keeping Meta AI mics. See what that changes for bystanders.

Can Meta Be Forced to Delete a Bystander's Data?

Not by this letter, and in the United States there is no general right for a bystander to demand it. The senators asked the question, "whether people can request their biometric and other data captured by its glasses be deleted," which signals they see a gap. Whether any answer exists depends on where you live and what the footage contains.

• EU and UK. The GDPR and UK GDPR give individuals rights of access and erasure against a data controller, and video of identifiable people is personal data. A bystander can in principle write to the controller, but must first work out who that is and whether the household exemption applies to the wearer. Our GDPR video compliance guide explains the controller/processor split in plain language.
• United States. A patchwork applies. California residents have deletion rights under the CCPA when a business holds their personal information (see CCPA and video privacy), and Illinois's BIPA covers face geometry. But a bystander often does not know that their face was captured, so rights go unused.

That asymmetry is the policy problem the letter points at. Rights that depend on the data subject finding out are weak against a camera that looks like ordinary eyewear.

What Are People Saying About the Letter?

At the time of writing there was no Hacker News discussion of the letter that we could link, so we are not summarizing community reaction. What can be said is that the phrase "pervert glasses" has become common shorthand for the harassment use case: 404 Media reports that guerrilla ad makers have produced anti-AI glasses ads in several countries and that people have started using the nickname in response to widespread invasions of privacy the glasses have come to represent. That is the outlet's characterization of public mood, not a poll.

What Does This Mean for Creators Who Film in Public?

If you publish video with strangers in it, you carry the risk the senators describe, whatever camera you use. Smart glasses make it easier to capture people without noticing, which raises the odds that a stranger's face, workplace or license plate ends up in your upload.

Practical rules:

• Assume bystanders did not consent. The fact that a clip was legal to record does not make it safe to publish, especially if the person is at work, in distress or in a sensitive place such as a clinic, gym or shelter.
• Remove what makes people findable. Faces matter, but so do name badges, license plates, shop signs, phone screens and voices.
• Think about the platform too. Where footage goes after upload, including into AI training, is governed by the service's terms. Read them before you rely on a default.
• Do not publish harassing content. Filming people who have not agreed to be on camera in order to provoke or embarrass them has led to doxxing, stalking and, in the reported cases, real harm to the people shown.

For a broader checklist, see our smart glasses content creation privacy guide.

How Can You Blur Bystanders in Smart Glasses Footage with BGBlur?

You can upload the clip to BGBlur in the browser, let the AI detect faces, and export a version where bystanders are blurred. This is the part of the problem that sits in your hands: what you publish.

1. Open BGBlur's face blur tool and upload your MP4, MOV or M4V file.
2. Run face detection. Motion tracking follows each face through the clip so you do not need to edit frame by frame.
3. Deselect anyone who agreed to be shown, such as you or a friend, and keep the blur on everyone else.
4. If cars are visible, add license plate blur as well. Our guide on why to blur license plates and faces in videos explains the reasons.
5. If voices can identify people, use voice anonymization on the audio.
6. Preview the whole clip, check edges and fast movement, then export.

Honest caveats. BGBlur does not stop anyone else from recording you, does not change what Meta's glasses capture, and cannot undo footage that already left a device. It works on files you upload and edit, and videos are deleted within 24 hours. The free plan has limits (200MB, 10 minutes, 3 videos per month, 720p export), and Pro removes the watermark and allows 1080p. Always review the output yourself; no automatic tool is perfect with small, side-on or partly covered faces. For why heavy blur rather than light pixelation matters, read about forensic de-pixelation and why weak blur fails.

What Should You Do Now?

Whether you wear the glasses, work around people who do, or publish street footage, a few steps are within reach today:

1. If you own Meta glasses, review the privacy and AI settings in the Meta AI app, turn off options to share media for improving AI where you can, and avoid using visual AI features around other people's intimate or sensitive moments.
2. Never disable the recording light and tell people nearby when you are filming. Many venues already ban camera glasses; check the rules.
3. If you publish clips, blur faces, plates and screens before upload, as in the steps above.
4. If you are filmed and posted without consent, save the URL, screenshot it, report it to the platform under its privacy or harassment policy, and ask for removal. Our guide on unauthorized filming and face blurring walks through options.
5. In the EU or UK, consider a data subject request to whoever published or controls the footage.
6. Watch for Meta's reply. If the company answers the senators publicly, the answers on retention, human review and deletion will tell bystanders far more than any marketing page.
7. Tell your representatives if you want bystander rules in federal law. The letter shows the issue has attention in the Senate.

Conclusion

The senators' letter does not change what Meta's glasses do today, but it names the issues clearly: a camera that looks like eyewear, bystanders who never agreed to be recorded, footage that may be retained and reviewed by humans for AI training, and no obvious way for the people in the frame to ask for deletion. The reporting behind it, from 404 Media's investigation to the lawsuits and the Swedish journalists' findings, is a mix of documented facts and allegations. Meta's answer will matter.

For anyone who publishes video, the lesson is practical. You cannot control other people's glasses, but you do control what appears in your own uploads. Blur faces and license plates, anonymize voices where needed, and review the result before you share. If you want to do that quickly, try BGBlur's face blur in your browser. We will update this post if Meta responds publicly or if the senators set further deadlines.

For related coverage, read our post on Ring's Familiar Faces feature and the lawsuit over doorbell face scanning, which raises the same bystander question for home cameras.