The truth behind the interview videos of women wearing offensive bikinis

The issue of AI garbage is becoming increasingly controversial. Photo: SCMP .

A recent spate of street interview videos featuring bikini-clad women has prompted lewd comments. The content is entirely AI-generated, designed to fool algorithms with numbers and draw attention with sexist imagery.

Realistic-looking videos are mass-produced by cheap AI tools, just by typing in a simple description command. Behind it is a small industry, using AI to create large numbers of videos, earning interactions to get paid by the platform.

Researchers at the French agency Agence France-Presse tracked hundreds of such videos on Instagram. Many of the situations, including changing rooms and scantily clad female reporters interviewing on the streets of India and the UK, have raised concerns about women’s safety.

Several of the videos, in Hindi, depict young men casually making sexist jokes and even touching women, with men in the background staring or laughing.

Many videos have attracted tens of millions of views, some of which have used their popularity to make money by promoting adult chat apps. Meanwhile, viewers have sometimes been confused, commenting and questioning the reality of the content they are watching. According to analysis by the US cybersecurity company GetReal Security, some of the videos were created using Veo 3, Google's realistic AI generator.

“This is part of the gendered harms caused by AI-mediated technology,” said Nirali Bhatia, a cyberpsychologist in India, who stressed that the trend is “feeding sexist ideology.”

Videos of women created by AI. Photo: Internet.

This trend also reflects an internet landscape increasingly polluted by AI-generated videos and images that compete for attention. Women have also become a tool for engagement. Many viral videos feature women exploiting men for money or sensationalizing the female aquatic trainer attacked by a killer whale.

Last year, Alexios Mantzarlis, director of the Security, Trust, and Safety Initiative at Cornell Tech, discovered 900 Instagram accounts that likely belonged to AI-generated “models.” The majority of the account holders were female and often dressed in revealing clothing.

Together, the fake accounts have attracted 13 million followers and posted more than 200,000 images, and the people behind them often make money by directing their audiences to commercial content sharing platforms.

Given the prevalence of AI fakes online, Mantzarlis says the numbers are likely to be much higher. “We’re going to see a lot more nonsense content that exploits body standards that are not just unrealistic but completely untrue,” he predicts.

On YouTube and TikTok, many creators have also opened paid courses, instructing how to monetize AI content. Many platforms like Meta have cut down on content moderation. In contrast, YouTube has been more aggressive and turned off monetization for AI videos.

“AI didn’t invent misogyny, it just mirrored and amplified what already existed,” said Divyendra Jadoun, an AI consultant. He believes that algorithms and AI proliferation will continue as long as audiences continue to interact with this content.

Comment (0)