Google launches and rapidly removes AI image generation feature from Google Earth after one day due to misinformation concerns
Feature update ● Confirmed 95% confidence first seen
Google introduced Nano Banana, an AI image generation feature in Google Earth that allowed users to create synthetic satellite imagery from text prompts. The tool was disabled within one day after users demonstrated it could easily generate false images of sensitive locations such as refugee camps, bombing sites, and nuclear facilities, prompting Google to acknowledge the misinformation risks and commit to reimplementing the feature with stronger safeguards.
Decision brief
- What changed
- Google launched an AI image-generation feature called Nano Banana (2) inside Google Earth on July 30, letting users generate synthetic satellite/aerial imagery from text prompts superimposed on real maps; it was pulled within a day (July 31) after users showed it could fabricate images of sensitive locations like refugee camps, bombed hospitals, and nuclear facilities.
- Why it matters
- Google Earth is widely trusted as a reference and open-source-intelligence tool, so embedding easy fake-imagery generation directly threatens the credibility of geospatial data used in journalism, security analysis, and due diligence. The episode shows how quickly a generative AI feature can be weaponized for disinformation even with watermarking (SynthID) in place, forcing companies to weigh launch speed against misuse risk and reputational fallout. Leaders overseeing AI product rollouts, brand trust, and information security should treat this as a concrete case study in insufficient pre-launch safeguards.
- Evidence
- Seven outlets—including The Verge, TechCrunch, 404 Media, and two Ars Technica pieces—independently and consistently report the July 30 launch, the July 31 removal, the Nano Banana 2 branding, and specific misuse examples (fake refugee camps, bombing sites, nuclear facilities), indicating strong cross-source corroboration.
- What remains uncertain
- It is unclear how many users generated misleading images during the roughly one-day live window, what specific technical or policy safeguards Google will add before relaunch, or on what timeline reimplementation will occur; reliance on SynthID watermarking as a mitigation is unproven against real-world misuse per Ars Technica's own critique.
- Monitor next
- Watch for Google's announced relaunch of the feature and what specific verification, geofencing, or approval safeguards it includes.
Analytical support, not advice — assumptions and open questions stated above.
Source coverage
Google's SynthID watermark is hard to break, but it doesn't solve AI misinformation
Ars Technica
independent
Google Earth launches Nano Banana image generation for realistic place redesign and historical reconstruction
The Neuron
newsletter
Here’s the problem with putting an AI image generator in Google Earth
The Verge
independent
Google Earth’s AI deepfake tool only lasted one day
The Verge
independent
Google Earth’s New AI Lets Anyone Fabricate Completely Bullshit Satellite Images
404 Media
Google nixes its Earth AI feature one day after launch, amid criticism it would spread misinformation
TechCrunch AI
independent
Google Earth risked ruin with retracted AI tool for making fake satellite pics
Ars Technica
independent