What Meta Promised and What a Review of Reels Found
In mid-July, Instagram head Adam Mosseri said the platform would remove harassing pickup-line and prank videos filmed with Meta camera-equipped glasses. The statement was unusually specific: Meta did not want people surreptitiously recording others, harassing them and then posting the footage on Instagram.
A month later, a Business Insider review found enforcement is still uneven. The reporter identified dozens of similar videos from popular and verified accounts. After Meta was sent more than 20 examples, it removed nine videos, including a clip in which a creator wearing Meta glasses asked a woman on the street if she was a secret code. Other videos from the same creator remained live.
Meta says thousands of pieces of content have been removed and several large accounts taken down. Search terms such as ‘rizz’ and ‘cold approach’ are now blocked on Instagram. Yet one large account was briefly reactivated after an earlier article, and a creator account deactivated after a journalist inquiry was restored the following day with the videos back up. TikTok and YouTube also removed some similar content after requests. The inconsistent pattern matters because Meta glasses sales topped 7 million units in 2025, more than triple the previous year, making the product a growing public presence.
Why Instagram’s AI-Glasses Moderation Problem Runs Deeper Than One Pledge
The Product-Reputation Risk Is Now Attached to Every Reel
These videos are not merely a moderation problem. They create a direct liability for Meta’s hardware bet. Chief Executive Mark Zuckerberg has said people without AI-capable glasses could face a significant cognitive disadvantage in the future. Sales growth and the smart-glasses race make the product a bright spot for Meta’s money-losing Reality Labs division. But the ‘creep glasses’ label and the flood of comments calling Kylie Jenner’s Meta glasses ‘pervert glasses’ show that online harassment content is shaping consumer perception faster than Meta’s own product messaging.
Why Enforcement Has Been Uneven
Meta’s Community Standards prohibit sexual harassment and the sexualization of adults without consent, but public-facing rules leave room for discretionary shifts, according to Sarah T. Roberts of the UCLA Center for Critical Internet Inquiry. The pattern described in the article suggests enforcement is often reactive: content is removed after a journalist sends links, then sometimes reappears through platform error or appeal. Creator Caren Babaknia, who has 174,000 followers, was deactivated after one of his Meta-glasses videos was reported, but his account was restored a day later with the videos intact.
The AI Moderation Promise Is Being Tested in Real Time
Meta told the Financial Times in June that initial tests showed its AI systems made fewer mistakes than human moderators and flagged more violating content. If that were already true at scale, the feed of a reporter engaging with pickup-line videos should not have become filled almost entirely with the same genre. The jury is still out on long-term AI performance. Mosseri’s direct personal statement is unique among video platforms because Instagram has a direct commercial stake in the hardware used to make the content, unlike TikTok and YouTube, which face the same genre without the same product exposure.
Three Unfinished Jobs for Meta’s Glasses, Trust and Ad Teams
For Meta, creators and advertisers, the review points to concrete gaps rather than a need for passive monitoring.
- Meta’s trust and safety team: Reconcile the June Financial Times claim that AI moderation outperforms humans with the finding that more than half of 20-plus flagged videos remained live a month after Mosseri’s mid-July pledge. Run an audit of Reels tagged Meta glasses, ‘rizz’ and ‘cold approach’ instead of relying on press-reported links to trigger removals.
- Meta’s hardware and communications teams: The report confirms that advertising, including a gambling app ad, appeared inside a Babaknia video, and that sales topped 7 million units in 2025. Clearer creator guidelines for glasses-filmed street content should be published before the next product cycle, because the ‘creep glasses’ and ‘pervert glasses’ labels now travel with the device name.
- Advertisers and brand-safety teams: Ask Meta for a current placement report on Reels adjacent to AI-glasses prank and pickup content. Use the Babaknia example, in which an account was deactivated after a journalist inquiry and restored the next day, as a concrete case when negotiating brand-safety guarantees for Meta placements.
Risk & Opportunity Assessment
| Commercial Risk | Medium | Sales of Meta glasses topped 7 million units in 2025, but harassment content is now directly tied to the product’s public image; sustained ‘creep glasses’ backlash could cool adoption. |
| Competitive Risk | Medium | Meta is the early leader in the smart glasses race, but the ongoing association with harassing content creates an opening for rival hardware makers if consumers hesitate. |
| Regulatory Risk | Low | The story involves Meta’s Community Standards and platform enforcement rather than a named regulatory proceeding, though uneven takedowns could attract formal scrutiny if the issue grows. |
| Reputation Risk | High | The ‘creep glasses’ label and the Kylie Jenner comment flood are already part of the product narrative; videos with millions of views continue to surface weeks after Mosseri’s pledge. |
| Technology Disruption | Medium | Meta is betting on AI to improve moderation, but the persistence of easily surfaced glasses-filmed harassment tests whether its moderation systems can handle new forms of this content at scale. |
| Commercial Opportunity | Medium | The same 2025 sales surge and first-mover position represent a significant opportunity if Meta can separate the glasses brand from viral harassment before competitors close the gap. |
Comments 0