AI Fakes Are Poisoning the One Corner of the Internet People Trusted Most: Animals
From lost-pet scams to faked rescue footage, AI-generated animal imagery is eroding public trust in real conservation work and breaking the hearts of owners still searching for their pets.
Key points
- A Los Angeles woman, Mibbby Butler, was sent at least one AI-generated fake photo of her missing cat Brooklyn in April 2025, part of an ongoing scam.
- Wired confirmed ChatGPT can recreate a convincing fake "found pet" photo using a 29-word prompt and a single image from a missing poster.
- The nonprofit We Animals, which has published work from 175 photojournalists documenting farm and lab conditions, now faces audience accusations that its real footage is AI-generated.
- New laws in California and the EU require AI image generators to embed invisible tags in synthetic content so social media platforms can label it.
- New York University researcher Jeff Sebo is pushing AI developers to add guidelines that discourage their models from producing content that could harm animals.
Somewhere between the polar bear rescue clips and the viral dairy-farm drone footage, a simple pleasure quietly broke. Scrolling past a cute or heartbreaking animal video used to feel safe. Now, for a growing number of people, it does not.
The explosion of AI-generated imagery over the past few years has turned animal content into one of the most polluted corners of social media. The damage is not abstract. It is landing on real people, real organisations, and real animals.
What happened to Mibbby Butler and her cat Brooklyn?
Butler's cat Brooklyn went missing in Los Angeles in April 2025, and a stranger quickly sent her a photo claiming to have found him. The picture looked convincing until she looked harder.
The new photo mirrored the missing poster she had posted online: same cat pose, same Torani syrup bottle in the background, same microwave. But the label text on the bottle was garbled, a classic giveaway that an image was built by an AI tool rather than a camera. Butler refused to pay the "temporary care" fee the stranger demanded. Brooklyn is still missing, four months later.
This was not a one-off. Butler told Wired she now receives a fresh deepfake, a fake photo generated by AI software, roughly once a month from different accounts claiming to have found her cat. Testing by Wired confirmed that ChatGPT generated at least one of those images. Using just 29 words and the original missing-poster photo, the publication recreated a near-identical fake in minutes.
Why does this hurt organisations trying to protect real animals?
Real evidence of animal suffering is now routinely dismissed as fake, even when it is not.
We Animals, a nonprofit that has published photojournalism from 175 photographers documenting conditions at farms, circuses and laboratories, has begun fielding accusations that its footage is AI-generated. A recent drone video showing rows of individual calf hutches on an Arizona dairy farm drew an Instagram comment this month asking simply: "How to prove it's not AI?"
The group bans its photographers from using AI tools entirely. Even so, it now plans to publish behind-the-scenes clips and adopt content-provenance technology, software that embeds the recording and editing history directly into a digital file, so viewers can verify what they are looking at. Victoria de Martigny, the group's director of visual content, said losing that trust could "open up the door to people questioning all of the work."
Fundraising is also at risk. Real, emotionally resonant animal footage is a primary tool for conservation and welfare organisations trying to attract donations. AI-generated content is cheap to mass-produce, and it is crowding out authentic material in social feeds and search results.
| Platform signal | What it means for viewers |
|---|---|
| Garbled text on labels or signs | Strong indicator of AI generation |
| EU / California invisible tags | Platforms must display an AI label |
| ChatGPT / Gemini image check | Can confirm if that tool made the image |
| Content-provenance file data | Records camera, location and edit history |
What can ordinary people actually do right now?
Verification tools exist, but they are patchy. ChatGPT, Google Gemini, and Meta AI can each check whether an image was made by their own systems, though none can catch fakes produced by a rival tool, and all impose limits on how many images a user can check per day.
The most practical steps right now are simple. If a "found pet" photo arrives with a payment demand attached, look for garbled text on any labels, signs or packaging in the image. That is the single fastest tell. Report the account to the platform. Butler did not report her scammer to the police because she was unsure a crime had occurred, which is an understandable reaction but one that lets the scam run.
For conservation content, the arrival of provenance-tagged files will help, though it will take time before the technology is standard.
Jeff Sebo, director of the Center for Mind, Ethics, and Policy at New York University, is asking AI developers directly to build guidelines into their models that discourage generating misleading animal content. The goal, he says, is keeping AI responses "grounded in evidence and reason" without making the tools refuse every reasonable request.
Legislation is moving too. California and the EU have both passed laws requiring the most widely used AI image generators to embed invisible markers in synthetic content. Social media platforms are then required to read those markers and display a visible label to users.
None of that helps Butler today. She spends more time now in a Facebook group of nearly 300,000 members who share only human-made art. It has become, she says, a reliable refuge while she waits for Brooklyn to come home.
Common questions
How can I tell if an animal photo is AI-generated?
Look at any text visible in the image, such as labels, signs or packaging. AI image tools frequently produce garbled or nonsense letters even when the rest of the picture looks polished. You can also upload a suspicious image to ChatGPT or Google Gemini, which can confirm whether their own system generated it, though neither tool can detect fakes made by a different AI.
Is it illegal to use AI to scam someone over a lost pet?
That depends on jurisdiction and how the scam is structured. Butler chose not to report her case because she was unsure a crime had occurred. If someone demands money using a fake photo, that can meet the legal threshold for fraud in many places. Reporting to local police and to the platform where the contact originated is the clearest first step.
