Instagram Showing Users Grotesque Videos of Human-Animal Hybrids

You may be pretty happy with the Instagram feed you’ve curated for yourself. But a recent episode on the app has served as a rude reminder to users that they aren’t in control of the mysterious algorithms they depend on for a quick screentime fix.

Kelly Takasu, a 36-year-old mom in Los Angeles, is one of those users. After she looked up videos about breastfeeding, Instagram took that innocent query as license to show her the weirdest AI slop imaginable.

“I noticed one day I had a Reels suggestion: an older woman who had a bunch of babies around her. I must have looked at it for too long,” she told the Washington Post.

And then followed more abominations: an AI-generated video of an older woman breastfeeding baby human-panda hybrids, and the piece de resistance, a video of women giving birth to other animal-human chimeras.

It was a pretty extreme departure from Takasu’s typical feed filled with parenting tips and cooking videos. At the very least, maybe she can take solace in the fact that she was far from alone.

In late February, Meta apologized for an “error” that led to many Instagram feeds being flooded with extreme content and gore, including shootings, deadly accidents, and videos of real-life murders. Outraged users took to social media, saying they felt borderline traumatized by being forced to see the graphic content.

Courtney, the parent of a 17-year-old boy in Virginia, recalled how it affected her son. 

“He saw assassinations — almost like terrorism. He saw a woman give birth, he saw just dead bodies, that kind of stuff,” Courtney told WaPo. “I feel like it was an attack of some sort on their minds. They’re still very impressionable.”

The intrusion underscores the blackbox nature of the algorithms controlling users’ feeds, and the never ending battle of policing extreme content on the internet. In Meta’s case, some graphic videos are flagged with a “Sensitive Content” screen, which blurs the footage and provides a warning before giving you an option to bypass it.

Still, that means that a lot of shocking content is allowed to run rampant on Instagram, under this pretty thin form of moderation. It’s worth mentioning that Meta recently loosened its content moderation policies, in a move that was widely criticized as essentially allowing hate speech and misinformation to go unpunished on its apps. The company has insisted, however, that the wave of disturbing content was unrelated to these changes, 404 Media reported.

Nonetheless, the shock-fest is also clear symptom of the deluge of AI slop inundating social media platforms with images that are uncanny, bizarre, or extracted from an alternate reality.

Because of their provocative nature, the AI images garner tons of engagement with relatively little effort, and can be churned out at a rate that no human moderation efforts could ever have any hope of keeping up with.

More on social media: Pinterest Changes User Terms So It Can Train AI on User Data and Photos, Regardless of When They Were Posted