AI-generated images of animals are now common enough on social feeds to break one of the internet's most reliable trust economies: cute-animal content. Pet owners chasing lost cats, nonprofits documenting farm abuse, and wildlife activists raising rescue funds are all reporting the same problem — audiences can no longer tell what is real, and scammers, slop farms, and skeptics are exploiting the gap.
The most direct harm is financial. Mibbby Butler, whose cat Brooklyn went missing in April after allegedly being dumped 30 minutes from her Los Angeles home, received a text with a photo of the cat sitting on a stranger's kitchen counter. The sender demanded money upfront for the cat's care. Butler noticed the Torani syrup bottle in the background had garbled label text — a classic generative-image tell — and refused to pay.
Four months later, Brooklyn is still missing, and Butler says she now gets a new deepfake of her cat from a different scammer about once a month. She did not report the original incident to police because she was unsure a crime had occurred.
“I didn't know people scammed for lost pets”— Mibbby Butler, Los Angeles pet owner targeted by an AI lost-cat scam
Key facts
- 01Mibbby Butler receives about one AI-generated deepfake of her missing cat Brooklyn every month from scammers demanding upfront payment.
- 02We Animals, which publishes 175 photojournalists documenting alleged animal abuse, plans to adopt provenance-embedding technology to defend real footage.
- 03WIRED reproduced a scam image nearly identical to one sent to Butler using a 29-word ChatGPT prompt and her original missing-pet photo.
- 04New laws in California and the EU now require major AI image generators to embed invisible tags marking synthetic content.
- 05A Facebook group of nearly 300,000 members for artists opposed to AI has become a refuge for users seeking human-made animal content.
The scale of the problem is measurable on the production side too. Using a 29-word prompt and Butler's original missing-pet poster image, WIRED was able to prompt ChatGPT to generate a picture nearly identical to one the scammer had sent. ChatGPT's own verification feature later confirmed it had generated at least one of the images Butler received. The barrier to producing a convincing fake pet photo is now a text box.
The trust damage extends past scams and into serious journalism. We Animals, a nonprofit that has published the work of 175 photojournalists documenting conditions at farms, circuses, and research labs, bans the photographers it works with from using AI. That does not stop viewers from accusing it of doing so anyway — an Instagram commenter on the group's drone footage of calf hutches at an Arizona dairy farm asked earlier this month, "How to prove it's not AI?"
Victoria de Martigny, We Animals' director of visual content, said the group now expects to publish behind-the-scenes clips and adopt provenance technology that embeds the origin and edit history of files directly into the media. Eva von Jagow, the group's marketing manager, said the shocking nature of some footage already invites skepticism, and AI has amplified that reflex.
Wildlife advocacy is facing the same drag. Oscar Horta, a philosopher and animal activist who recently co-directed a short film on AI's effect on wildlife, said fabricated clips of animals being pulled from floods and fires are crowding out real rescue footage in social feeds and search results. His concern is not just aesthetic — legitimate conservation groups depend on evocative real imagery to raise money, and viewers who have been trained to assume everything is fake stop donating. Horta called AI-generated scenes of sailors saving drowning polar bears "ridiculous" and "unrepresentative of what it means to help animals."
Researchers are starting to push back at the model layer. Jeff Sebo, director of the Center for Mind, Ethics, and Policy at New York University, is asking AI developers to update their model guidelines so systems are less willing to produce content that could mislead people about animal suffering or fabricate rescue scenarios.
Regulation is moving in parallel. New laws in California and the EU now require the most widely used AI image generators to embed invisible tags marking their output as synthetic, and require social platforms to surface that labeling to users. In-product verification already exists in ChatGPT, Gemini, and Meta AI — each can identify whether an image was generated by its own system — but the tools have per-user caps and require the viewer to upload each image manually. Butler's ordeal might have ended at the first scam text if her iPhone had flagged the image automatically.
The counterweight is that provenance tagging only works when every major generator complies and every major platform reads the tags, and the current legal reach covers California and the EU rather than the global feed. Bad actors can also route around labels using older or open-weight models that do not embed C2PA-style signals. The Facebook group of nearly 300,000 members for artists who oppose AI, where Butler now spends time to see human-made work, is a symptom of how far trust has slipped before the technical fixes have shipped.
For the AI industry, the animal-content problem is a preview of a broader authenticity crisis coming for every category of user-generated media. Provenance standards, in-app verification, and platform-level labeling are all technically viable today; the constraint is deployment speed and cross-vendor coordination. The vendors who build verification into the default consumer flow — not behind an upload step or a paid tier — will own the trust layer, and the ones that do not will end up as the tools scammers reach for first. Butler is still looking for Brooklyn.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
