Google And Meta Shipped AI Tools That Fake Evidence -- Then Pulled Them Days Later

Google and Meta have pulled AI tools that could generate fake evidence—including location-grounded synthetic images—after demonstrating that the accessibility and visual credibility of manipulation had outpaced verification workflows. The rollback removed a specific Google Earth feature, but the broader editing capability remains in Gemini, where users can upload images such as Earth or map screenshots and request edits through natural language. The article says Google’s generated images were watermarked as AI-made, yet verification tests using Google’s SynthID pathway returned inconclusive results for a high-risk fake image. The concern is not that all authentic imagery is suspect, but that authentication can be undermined quickly and at scale.