Google Pulls AI Image Generation from Google Earth After Just 24 Hours
Users mass-created deceptive scenes, including 9/11 attacks and fabricated protests, prompting Alphabet to disable the tool until stronger safeguards are in place.
Users mass-created deceptive scenes, including 9/11 attacks and fabricated protests, prompting Alphabet to disable the tool until stronger safeguards are in place.
Alphabet's Google announced on July 31, 2026, that it is withdrawing the AI image generation feature it introduced in Google Earth just a day earlier. The decision came after users began sharing generated images that violate the company's policies, as reported by The Straits Times, citing Reuters.
The feature, powered by Google's AI model Nano Banana 2, was intended for professionals and allowed users to generate photorealistic images based on Google Earth's satellite, aerial, and 3D imagery by entering a text prompt. However, shortly after launch, disturbing creations began circulating online. According to Ukraine's UNIAN, users added "tons of madness" to maps: a plane from the September 11, 2001 attacks near skyscrapers, nonexistent explosion craters next to landmarks, and fabricated protests.
Google acknowledged the issue via X (formerly Twitter). "We've seen geospatial experts using this feature for a range of useful purposes, but we've also seen people sharing generated images that appear to violate our policies," the company told The Straits Times, adding that they are pausing the capability while they work on stronger safeguards. UNIAN reports that Google admitted such images undermine trust in the service as a reliable source of information about the world. Google did not specify exactly which types of images violated the rules but emphasized that they did not appear in the main Google Earth view and were watermarked as AI-generated content.
This incident is not isolated. Tech giants are rushing to integrate AI tools into their products, but they are often forced to adjust, restrict, or temporarily disable features due to misuse and concerns about misinformation. The Straits Times recalls that Meta earlier in July 2026 shut down its AI feature Muse Image, which allowed users to generate images from public Instagram accounts. That move followed widespread criticism, including from Hollywood unions, over privacy concerns.
On the same day the Earth tool was pulled, Google also announced a range of other AI updates on its channels. The AI agent Gemini Spark began rolling out to subscribers of Google AI Pro outside the US, and soon after to those in the US. Spark is designed as a personal agent that works in the background 24/7. Additionally, an interview with the Google DeepMind Robotics team about the latest Gemini Robotics 2 model was released, along with the monthly Gemini Drops update featuring new capabilities for desktop and mobile devices.
While the news has no direct connection to Croatia, it globally illustrates the challenges tech companies face when implementing generative AI technology, particularly when it comes to maintaining trust in services considered reliable sources of information about the world.
