Since Israel and the United States began striking Iran in February, the Gulf region has become a battlefield for both kinetic attacks and information warfare. Analysts, journalists and humanitarian groups have turned to commercial satellite sources, especially Google Earth, which for two decades has been treated as a reliable “ground truth” for monitoring sites in Syria, Yemen and Sudan.
On July 30 Google added a generative AI layer to Earth, allowing users to zoom to exact coordinates, type a natural-language prompt and receive a photorealistic image blended with real satellite data. Within hours the community produced a fabricated nuclear plant in Iran, a bomb crater beside a Gaza hospital and floodwaters surrounding the U.S. Capitol, all appearing as seamless extensions of the original map.
Investigative experts and media outlets quickly condemned the experiment, prompting Google to withdraw the feature the next day and promise “stronger guardrails.” At the same time the U.S. government ordered satellite firms to limit journalists’ access to imagery in the theater, further shrinking the pool of authentic visual evidence while synthetic content spread.
The episode mirrors other recent AI rollbacks. OpenAI discontinued its Sora video service in March after public pressure over abusive outputs, and Google’s own Veo 3 video generator, released in May 2025, produced watermarked missile-strike clips that users later cropped out. An unwatermarked AI video of a blast at Iran’s Evin prison circulated on Sky News and the BBC before Citizen Lab identified it as likely part of an Israeli-run influence operation.
Regulatory frameworks have lagged behind the technology. The EU AI Act’s transparency duties began on August 2 2026 but remain voluntary, while California’s AI disclosure law, effective August 2024 for generative-AI creators, will not apply to platforms until at least 2027. Tests of the mandated detection tools showed that seven of thirteen providers lacked public detectors and the remaining tools could be easily deceived.
Legal scholars Bobby Chesney and Danielle Citron coined the phrase “the liar’s dividend” in 2019 to warn that convincing fakes enable bad actors to cast doubt on genuine evidence. The Google Earth incident demonstrated how a single day of fabricated imagery can erode confidence in verification pipelines, underscoring calls for mandatory provenance tags, robust detection mechanisms and a burden of proof that rests on the companies deploying such systems.