AI
Google pulls Earth AI image generator after 24 hours over misinformation fears
The company disabled a feature that let users overlay fictional images onto satellite maps, citing abuse within hours of launch.
TechCrunch — AI
Google has withdrawn an AI image generation feature from Google Earth just one day after introducing it, following swift pushback about potential misuse.
What happened#
The search giant launched the capability Thursday that integrated its Nano Banana 2 image generator into the mapping service, permitting users to generate and place fictional images directly onto real satellite imagery. Google framed the tool as enabling creative geographic exploration.
Why it faced backlash#
The feature drew rapid criticism from journalists, researchers, and digital-integrity advocates who flagged a fundamental risk: layering artificial imagery onto maps widely trusted for factual accuracy could readily fuel misinformation. Because the tool accepted virtually any text prompt and composited the results onto authentic geographic data, observers warned it would enable geospatial disinformation at scale.
"There's no way that this new AI image generation feature on Google Earth, one of the most reliable sources of visual evidence for journalists and researchers, could possibly be abused to spread misinformation online," one BBC journalist noted sarcastically on social media.
Google's response#
Within hours, the company moved to deactivate the feature. In a statement, Google acknowledged observing both legitimate use cases among geospatial professionals and misuse—specifically screenshots of generated images breaching its policies. The company said it would "roll back this feature in Google Earth while we work on implementing stronger guardrails."
The broader context#
The episode highlights a recurring tension in the AI era: visual media online can now be fabricated or altered with minimal technical skill, and verified sources can become vectors for falsehoods if they host generation tools. While critics' concerns are valid, the underlying challenge extends beyond this single feature—most imagery circulating digitally is now vulnerable to manipulation.
