The Fake Baby That Fooled the Internet

Zendaya isn't pregnant. AI invented the moment. Nobody caught it in time.

The Photo That Broke the Internet

Last week, a photo of Zendaya and Tom Holland walking through a park went viral. She appeared to have a visible baby bump. Within hours, "Zendaya pregnant" was trending on X, Instagram and TikTok. Fan accounts were celebrating. Tabloids were running with it.

The photo was fake. An AI tool had taken a real image from a 2023 London street, quietly adjusted the shape of her stomach, and released it into the world. The actual picture shows Zendaya with a flat belly. Nobody had spotted the swap.

This Has Happened Before

This is not a one-off. Earlier this year, AI-generated wedding photos of the couple did the same rounds. Zendaya went on Jimmy Kimmel Live and said "Many people have been fooled by them." An insider told reporters there is "a whole parallel universe of fake moments being created" for celebrities. Not satire. Not art. Just manufactured reality, distributed at scale, designed to be believed.

The Tell Is Gone

The thing worth sitting with is this: the images are getting better. The tells that used to expose AI fakes, weird hands, blurry backgrounds, inhuman lighting, are disappearing. The tools are cheap. The audiences are large. And the verification infrastructure, the systems and habits that help people work out what is real, has not kept pace.

What It Actually Means

Zendaya's team will move on. The rumours will die down. But the pattern here is the one that matters. AI is not just generating text or answering questions. It is now manufacturing celebrity moments, relationships, pregnancies, weddings, that never happened. Because these moments are built around real people with real emotional stakes for their fans, the misinformation spreads faster and lands harder than most.

This Week's Other Story

OpenAI launched a teen-specific version of ChatGPT, automatically routing users aged 13 to 17 into a restricted mode that blocks romantic language, limits sensitive topics and adds study prompts. It is years after most teenagers started using it. The gap between when technology gets deployed and when the safeguards arrive is a recurring theme. Not a criticism, just something to notice.

The Question

If you cannot trust a photo of a celebrity walking in a park, what does that do to trust more broadly? Not just in media, but in the everyday evidence you rely on? That is not a rhetorical question. It is a design problem, and nobody has solved it yet.