Google Earth retracts AI editing tool after one day amid fake image fears

Google Earth's attempt to integrate AI-powered image generation into its platform was short-lived, as the company quickly retracted the feature amid widespread concerns about its potential to create misinformation. The tool, announced on July 30, would have allowed users to generate images based on satellite, aerial, and 3D map data, promising applications in historical visualization and planning.

However, experts and media quickly pointed out the significant risks. The Financial Times reported that fake images of conflict zones like Gaza and sensitive sites such as Iranian nuclear facilities were already appearing on social media. AFP's own tests confirmed the ability to create fabricated satellite images of politically sensitive events, including explosions in Paris and training camps of the Islamic State group.

“

We know the unique trust people place in Google Earth to show the world in a reliable way. We are rolling back this feature in Google Earth as we implement stronger safety measures.

Brian Horwitz, Google Earth's product manager, acknowledged the concerns, stating that while the feature had potential for useful applications, the company had also seen generated images shared that violated their policies. "We know the unique trust people place in Google Earth to show the world in a reliable way," Horwitz said in a statement. "We are rolling back this feature in Google Earth as we implement stronger safety measures."

Sam Stockwell, a senior researcher at The Alan Turing Institute, emphasized the damage such a tool could inflict on the integrity of visual evidence. "Google Earth has served as a reference layer for journalists and investigators for decades to cross-check dubious images," Stockwell told the Financial Times. "Putting a generative tool inside that benchmark pollutes the very yardstick used to expose fakes."

“

Google Earth has served as a reference layer for journalists and investigators for decades to cross-check dubious images. Putting a generative tool inside that benchmark pollutes the very yardstick used to expose fakes.