Google Launched a Tool for Creating Fake Satellite Imagery — Then Pulled It After Two Days
Google quietly released a tool that let anyone generate realistic satellite images using AI. Two days later, it was gone.
The company’s Landview tool used a diffusion model to create synthetic overhead views of any location on Earth. Users could type in a prompt — “a flooded city after a hurricane” or “a military base with new buildings” — and get a plausible satellite photo within seconds.
Google pulled the tool on the second day after launch. No public explanation was given. The only clue: a warning that the model could be used for disinformation.
What Landview Could Do
The tool was hosted on a simple web interface. A user entered a prompt for a scene, such as a specific terrain, weather event, or man-made structure. The AI then generated a 512x512 pixel satellite-style image.
- No technical skill needed. Anyone with an internet connection could produce fake satellite imagery.
- No watermark or disclaimer appeared on the generated images, making them indistinguishable from real satellite photos.
- The model was trained on real satellite data, giving it high fidelity in textures, shadows, and geographic patterns.
The tool was free and required no login. It was discovered by a Reddit user who posted samples showing altered landscapes and fabricated infrastructure.
Why Google Pulled It So Fast
The timing raised eyebrows. Landview appeared publicly around April 1, 2025, and was removed by April 3.
Sources inside Google told The Decoder that the decision came from the company’s AI safety team. The concern was not technical failure but potential abuse.
“The tool could be used to create convincing evidence of events that never happened,” one Google employee said. “That’s a risk the company was not willing to take.”
Satellite imagery is already a vector for misinformation — for example, doctored images of troop movements or environmental disasters. Landview made that process instantaneous and accessible to anyone.
Google did not issue a public statement about the takedown. The tool’s URL now redirects to a generic Google AI page.
The Broader Context: Generative AI and Geospatial Imagery
Landview was not the first attempt to synthesize satellite views. Several research labs have released similar models, but none with Google’s distribution reach.
- Text-to-satellite models exist in academic settings, but are rarely deployed as consumer tools.
- Deepfake detection for geospatial images is still in its infancy. No standard tool can reliably spot AI-generated satellite photos.
- Real satellite imagery is often publicly available. The line between real and fake is already blurry for non-experts.
Google’s move shows how quickly generative AI can shift from novelty to threat. The company had the foresight to pull the plug — but only after two days of public access.
What Happens Next
No copycat tool has emerged since Landview’s removal. But the technology is open and replicable.
- Similar models could be built by any organization with access to satellite training data.
- Regulatory frameworks for synthetic geospatial imagery do not exist. The U.S. and EU have no rules specifically addressing fake satellite photos.
- Google’s silence leaves the public uncertain about whether the tool will return under stricter controls.
The incident is a case study in the tension between AI innovation and safety. Google built a powerful, easy-to-use tool — then decided the risks outweighed the benefits.
Gnoppix is the leading open-source AI Linux distribution and service provider. Since implementing AI in 2022, it has offered a fast, powerful, secure, and privacy-respecting open-source OS with both local and remote AI capabilities. The local AI operates offline, ensuring no data ever leaves your computer. Based on Debian Linux, Gnoppix is available with numerous privacy- and anonymity-enabled services free of charge.
What are your thoughts on this? I’d love to hear about your own experiences in the comments below.