Google Halts AI Imagery Feature in Google Earth Following Disinformation Concerns
In a significant retreat from its recent technological expansion, Google announced on July 31 that it would suspend a newly launched feature for Google Earth that allowed users to generate AI-enhanced visualisations atop satellite imagery. The move came less than 24 hours after the rollout, following an immediate and intense backlash from the open-source intelligence (OSINT) community and disinformation researchers. These experts argued that integrating generative AI into a platform widely considered the gold standard for geographical truth posed an unacceptable risk to the global information ecosystem.
The feature, which was powered by Google’s “Nano Banana 2” image-generation model, was initially framed as an innovative tool for urban planning and historical visualization. By allowing users to manipulate satellite, aerial, and 3D-mapping data to create new visuals in seconds, Google hoped to provide a dynamic way for people to interact with the world. However, the company’s decision to pull the tool was swift once it became clear that users were utilizing it to create fabricated scenes that were indistinguishable from genuine intelligence imagery, directly violating the company’s stated safety policies.
Prominent researchers were vocal in their condemnation, warning that the tool fundamentally undermined the utility of Google Earth as a verification resource. Digital investigations expert Henk van Ess captured the sentiment of many critics, noting that Google had spent two decades establishing a baseline of truth, only to introduce a mechanism that could manufacture falsehoods. Brady Africk, an analyst at the American Enterprise Institute, echoed these concerns, stating that the ease of generating convincing fake satellite imagery threatened to erode public trust in geospatial data, thereby complicating the vital work performed by journalists and researchers tasked with holding governments and actors accountable.
As the controversy gained momentum, Google initially attempted to mitigate concerns by pointing to “SynthID,” a system that embeds invisible digital watermarks into AI-generated media to signal its artificial origin. The company emphasized its commitment to preventing misinformation and misleading activities, linking to its existing AI policies. However, reality tests conducted by news agencies—including AFP—demonstrated that the safeguard was insufficient to prevent the creation of potentially incendiary fabrications. The tool was successfully used to generate realistic, yet entirely fake, imagery of an explosion in Paris, nuclear sites in Iran, and military activity in sensitive regions, proving that the technology’s potential for harm outweighed its practical utility.
The potential for real-world consequences was a driving factor in the rapid mobilization of advocacy groups. Organizations like the Centre for Information Resilience and the volunteer-led OSINT project GeoConfirmed were instrumental in demanding the feature’s removal. They documented instances of individuals already using the tool to relabel buildings and manufacture false military movements, warning that the tool was being weaponized to manipulate the global information environment. These experts argued that trust in satellite imagery, an asset that has taken decades to build, could be “irrevocably damaged overnight” if such tools remained in the hands of bad actors.
Ultimately, Google’s decision to roll back the feature serves as a stark case study in the risks of integrating generative AI into platforms that the public relies on for objective truth. By acknowledging that people “uniquely trust Google Earth for a reliable view of the world,” the company admitted that the feature was fundamentally incompatible with the platform’s core identity. While Google has stated it will work on “stronger guardrails” before potentially reintroducing the technology, the incident marks a cooling period for AI integration in sensitive sectors, highlighting the persistent tension between rapid technological innovation and the preservation of verifiable reality.