Google’s new artificial intelligence feature for Google Earth has created a source-verification problem for open-source investigators, policymakers and military analysts who use maps and commercial imagery to corroborate events.
Google announced July 30 that its Nano Banana 2 feature allows web users to select a location in Google Earth and generate custom images from the platform’s satellite, aerial and 3D imagery [1]. The company says the feature is available globally and promotes uses such as visualizing historical scenes, real estate plans and proposed construction [1].
Open-source investigator Henk van Ess reported that he tested prompts that placed refugees near the U.S.-Mexico border, a nuclear plant in Iran and a hospital with a bomb crater in Gaza [2]. In a response quoted in his report, Google said generated images include SynthID digital watermarks and that it blocks image creation on harmful topics. Van Ess wrote that the prompts he tested were not refused [2].
“Yeah this is bad,” the defense analysis account OSINTtechnical wrote in response. The account said the feature did “immediate, catastrophic damage to the credibility of satellite imagery” [3].
The central concern is not that the feature changes what a satellite collected. It is that a synthetic image can inherit the visual authority of a familiar mapping environment after it is removed from that environment and circulated without collection metadata, provenance or context.
That distinction matters for tactical operations. The development does not demonstrate a loss of space-based surveillance capability. It creates a problem of imagery provenance that can affect the analysts, commanders and public audiences trying to interpret imagery after it enters the information environment.
For tactical formations, space dependence also extends beyond surveillance. The Army says positioning, navigation and timing data enables forces to move, shoot and communicate precisely, extend operational reach, control tempo and perform mission command [4]. Its current space-operations guidance lists assured PNT, communications, ISR and targeting among the capabilities integrated for land operations [5].
Satellite communications and PNT therefore remain central tactical dependencies, while imagery serves a separate and essential intelligence and targeting function. The new Google Earth feature does not erase that distinction. It makes it more important to establish what sensor collected an image, when it was collected and whether a second independent source corroborates its content.
For defense policymakers, the immediate lesson is about trust rather than orbital capacity. As generative tools make location-specific visuals easier to create, commercial imagery and open-source material will require stronger provenance checks before they inform public claims or operational assessments.
AI generated
References
[1] Google: Transform any place with Nano Banana in Google Earth
[2] Digital Digging: How to plant a nuclear plant in Iran
[3] X: OSINTtechnical
[4] U.S. Army: SMDC takes lead in Army’s navigation warfare operations
[5] U.S. Army: Space Operations (CMF 40) Career Progression