Google Earth’s AI Image Generation Function Pulled in Less Than a Day

Google pulled a brand new AI image-editing characteristic from Google Earth lower than 24 hours after launch.

The corporate launched the perform on Thursday, permitting Gemini’s Nano Banana 2 picture mannequin to entry Google Earth and rework areas with textual content prompts.

The perform, nevertheless, rapidly received out of hand. AI researcher Henk van Ess, for instance, created extraordinarily sensible deepfakes, together with pretend refugee camps alongside the US-Mexico border and a nuclear plant in Iran.

Different customers used the instrument to create AI pictures of a nonexistent flood and a 9/11-style assault.

By Friday, Google had reversed course.

“We have seen geospatial professionals utilizing this characteristic for a variety of helpful functions, nevertheless we have additionally seen individuals sharing screenshots of generated imagery that seem to violate our insurance policies,” a Google spokesperson mentioned in a press release posted on X. “So we’re rolling again this characteristic in Google Earth whereas we work on implementing stronger guardrails.”

Google added that the AI-generated pictures had been watermarked and by no means appeared within the public Google Earth expertise.

Google shouldn’t be the one firm that has these days rolled out AI options solely to rapidly take away them. Meta launched Muse Image on July 7 with a characteristic that allowed customers to generate AI pictures utilizing pictures obtainable on all adult-owned public Instagram accounts that had not modified their settings to decide out. Meta eliminated the characteristic in lower than per week and mentioned it had “missed the mark.”