Photographer wins AI contest with a real photo

For the past two years, the art world has been panicking over AI images sneaking into traditional photo contests and stealing awards. Now, a photographer just flipped the script.

Miles Astray entered a completely real, camera-shot photograph of a headless-looking flamingo into the AI category of the 1839 Color Photography Awards. Not only did the judges award it third place, but it also won the People’s Vote Award before organizers discovered the truth and disqualified him.

A glitch in nature fooled the pros

The image captures a flamingo tucking its neck in to scratch itself, creating a bizarre, surreal silhouette that looks suspiciously like a generative AI failure.

The 1839 Color Photography contest was evaluated by heavyweights from The New York Times, Getty Images, Christie’s, Phaidon Press, and Maddox Gallery. Not a single judge realized the shot was captured with a camera rather than a text prompt.

Astray told PetaPixel that his stunt was a deliberate experiment. After watching AI image makers fool prestigious competitions—most notably Boris Eldagsen taking a prize at the 2023 Sony World Photography Awards—Astray wanted to turn the table. He set out to prove that nature's strangeness can still outshine synthetic outputs, and that Mother Nature remains the ultimate creative force.

Organizers disqualified the entry, acknowledging Astray's powerful statement but pointing out that keeping the award would be unfair to genuine AI artists.

Why it matters

This is brilliant performance art that exposes an uncomfortable truth: our visual gatekeepers can no longer tell what is real and what is synthetic.

When senior photo editors from top-tier publications cannot distinguish a wild bird from a diffusion model render, it confirms that visual evaluation based solely on the final pixels has completely broken down. While Eldagsen showed that photography contests could not catch AI, Astray showed that AI contests cannot catch reality.

We think this highlights an interesting silver lining for human photographers. The world still produces random, organic oddities that machine learning models—trained on averages and predictable aesthetics—rarely imagine on their own.

However, it also proves that splitting creative contests by toolchain is becoming unsustainable. Without cryptographic origin tracking or camera-level C2PA credentials, juries are essentially guessing. Astray got his point across, but the industry now faces the headache of having to verify reality just as much as artificiality.