Stop trusting the animal videos in your feed [2026]

Mibbby Butler thought she’d found her cat. In April, a stranger texted her a photo of Brooklyn sitting on a kitchen counter, wrapped in a girl’s arms, four days after her roommate’s boyfriend allegedly took him and dumped him 30 minutes away in the Los Angeles suburbs.

Then came the ask: money up front, for Brooklyn’s supposed temporary care.

So Butler looked harder at the picture. Brooklyn was posing exactly as he had in the photo she’d used on the missing poster she’d posted to social media. Same Torani syrup bottle in the background. Same microwave. But the label text on the bottle was garbled into nonsense, which is the oldest AI tell there is.

She didn’t pay. Four months later, Brooklyn is still missing.

The scam that runs on grief

“I didn’t know people scammed for lost pets,” Butler says. She didn’t report it to police either, because she wasn’t sure a crime had actually happened.

About once a month, she gets another deepfake from someone new claiming to have found her cat. In tests WIRED conducted, ChatGPT confirmed it had generated at least one of those images, the one showing a purported Brooklyn eating in front of a doorstep. Then, using the original missing-poster photo and a 29-word prompt, WIRED produced a picture nearly identical to the scammer’s.

Twenty-nine words. That’s the whole barrier to entry for weaponizing someone’s lost pet against them.

Now the real footage is under suspicion too

The nonprofit We Animals has published work from 175 photojournalists documenting alleged abuse and other harms at farms, circuses and scientific labs. The material is shocking by design, which has always invited skepticism, says Eva von Jagow, the group’s marketing manager.

Drone footage the group shared of rows and rows of hutches allegedly used to shelter calves separated from their mothers on an Arizona dairy farm is a good example. The composition is so relentlessly geometric it reads as synthetic.

People used to accuse We Animals of staging photos or running them through Photoshop. Now they accuse it of using AI, even though the group bans its photographers from touching the technology. “How to prove it’s not AI?” one Instagram user commented on the dairy farm video earlier this month.

That’s the trap. There’s no clean way to prove a negative to an audience that’s been burned enough times to stop extending the benefit of the doubt.

What it costs to defend yourself

We Animals expects it’ll have to start publishing behind-the-scenes clips and spelling out how staffers verify submissions. It also plans to adopt technology that embeds provenance and editing history directly into the digital files.

Victoria de Martigny, the group’s director of visual content, says protecting that trust matters because losing it would “open up the door to people questioning all of the work,” and “we don’t ever want to be in that position.”

Think about the asymmetry. A photojournalist travels to a farm, gets access, shoots the footage, then builds a second layer of documentation just to prove the first layer happened. Meanwhile someone with a chatbot subscription generates a fake rescue in under a minute and collects the ad revenue.

Cheap fakes crowd out expensive truth

Real, emotionally charged images are how conservation and welfare projects bring in donations. For the people running AI slop accounts, synthetic imagery is just a shortcut to likes and ad money.

Because fakes cost so much less to make, they appear to be drowning out genuine clips in social feeds and search results, says Oscar Horta, a philosopher and leading animal activist who recently helped direct a short film about how AI could affect wildlife.

Horta says the far-fetched rescue videos worry him most: wild animals plucked from fires and floods, that kind of thing. He thinks the likely AI-generated content will push people to doubt legitimate rescue tactics and undercut fundraising for them later. Which is a bad trade at a moment when extreme weather events are becoming more common.

There’s a physical risk too. Fake footage can teach people to do dangerous things. “There are deepfakes of polar bears drowning and people on boats coming and rescuing them,” Horta says, calling the scenes “ridiculous” and “unrepresentative of what it means to help animals.”

The fixes on the table are partial

This month, Jeff Sebo, director of the Center for Mind, Ethics, and Policy at New York University, started pressing AI developers to write language into their model guidelines that would steer the systems away from generating responses potentially harmful to animals.

The goal, Sebo says, is to “emphasize the importance of staying grounded in evidence and reason” around the possible suffering of individual creatures, without being “overly preachy, overly moralizing, or refusing reasonable user requests.”

New laws in California and the EU now require the most popular AI image generators to embed invisible tags marking content as machine-made. Social platforms then have to read those tags and label the pictures and videos publicly.

ChatGPT, Gemini and Meta AI each have features that can tell you whether an image came from that particular chatbot. Useful in theory. In practice, nobody’s uploading every animal photo they scroll past, and there are limits, including caps on how many images you can check.

Building verification into messaging apps and browsers, switched on by default and with privacy protections, would close most of that gap. None of it exists yet.

Where she goes instead

Butler’s iPhone could have flagged that first photo as probably fake before she let herself believe it. It didn’t, and there’s no setting she could have changed.

She blocks every account that puts an AI cat video in front of her. New ones keep appearing.

So she’s been spending more time in a Facebook group for artists who oppose AI, roughly 300,000 members strong, where she can look at work she knows a person made. That’s the sanctuary now. A walled-off corner of the same platform, where she waits for Brooklyn.