Google Earth's new AI image generation function didn't survive a day after users deepfaked disasters (www.businessinsider.com)

Google has retracted its newly launched AI image-editing feature from Google Earth less than 24 hours after its debut, following widespread misuse by users creating highly realistic deepfakes of disasters. The function utilized Gemini's Nano Banana 2 model to transform geographical locations based on user prompts, but quickly led to creations such as fake refugee camps along the US-Mexico border and non-existent nuclear facilities in Iran. Prominent figures like AI researcher Henk van Ess highlighted the tool's potential for harm by generating implausible yet alarming imagery with ease. This incident underscores the challenges of implementing AI tools in public applications without sufficient safeguards. While Google acknowledged the feature's beneficial usage by geospatial professionals, the rapid generation of distressing imagery prompted immediate action to retract the feature. Google stated it is working on stronger guidelines to prevent misuse, emphasizing the importance of responsible AI deployment. This situation mirrors similar instances in the tech industry, where companies like Meta have also withdrawn newly released AI features due to ethical concerns.