Rest of World · Technology

Google Earth's AI Image Feature Pulled After 24 Hours Amid Trust Concerns

Google briefly integrated AI image generation into Google Earth, allowing users to create photorealistic scenes at real coordinates. The feature was quickly removed following widespread criticism from experts and journalists.

The new feature allowed users to type prompts and generate AI images blended with actual satellite imagery, creating fabricated scenes like a nuclear plant in Iran and a bomb crater near a Gaza hospital.

Google Earth has been a trusted tool for open-source investigators. The ease of creating fakes within this platform raised concerns about its reliability as a source of "ground truth."

The feature was pulled within a day after an outcry, with Google stating it would return with "stronger guardrails." This highlights a pattern of new AI fabrication tools launching before adequate safeguards are in place.

This follows similar issues with other generative AI products, like OpenAI's Sora video app. The article notes a recurring problem where harm occurs before companies implement necessary protections.

The U.S. government restricting journalists' access to satellite imagery in conflict zones further limits real information while AI-generated fakes proliferate.

The rapid spread of AI-generated content makes detection difficult. Forensic analysis often lags behind, allowing doubt to undermine authentic evidence, a phenomenon termed the "liar's dividend."

Experts call for mandatory provenance tracking and robust AI detection tools. The doubt cast by the Google Earth incident cannot be easily recalled, impacting the visual record.

AI-samenvatting op basis van de bron.