One of the most trusted sources of visual evidence for journalists and researchers just spent 24 hours being a misinformation machine.

Geospatial credibility was the casualty here

Fact-checkers and conflict reporters rely on satellite imagery as ground truth — verifiable, unedited, timestamped. Dropping a prompt-based image generator on top of that workflow does not add creativity; it corrupts the evidentiary chain entirely.

Google shipped this, watched it break, and reversed course in a single news cycle

Google nixes its Earth AI feature one day after launch, amid criticism it would spread misinformation integrated Nano Banana 2, Google’s AI image generator, directly into Google Earth, allowing users to superimpose AI-generated imagery over real satellite maps using text prompts. The input was a location and a prompt; the output was a fabricated but visually credible overlay on verified geographic data. Screenshots began circulating almost immediately, and Google confirmed the images violated its own policies before pulling the feature entirely.

The people who felt this fastest work with maps as evidence

• Investigative journalists who use satellite imagery to verify troop movements, disaster zones, or infrastructure claims — and now have to source-check every screenshotted map they receive.
• Geospatial analysts at NGOs or research institutions who treat Google Earth exports as citable visual records.
• Legal and insurance professionals who depend on unaltered aerial imagery to document property conditions or incident scenes.

The feature was live long enough for the trust damage to begin, even if the tool is gone.

Google pulled it, but the underlying problem is not going anywhere

Every major AI image generator — including products from OpenAI, Adobe, and Google itself — can already produce satellite-style imagery without any mapping integration at all. The real issue is not one feature rollback; it is that geospatial visual evidence now requires provenance verification that the industry has not built yet.

What this episode actually demonstrated

• Generate convincing fake satellite imagery without any technical skill.
• Embed fabricated geographic overlays in otherwise authentic-looking maps.
• Circulate screenshots that strip the AI-generation context entirely.
• Exploit trusted brand associations to lower a viewer’s skepticism threshold.

Google has not announced a timeline for re-releasing the feature with guardrails in place.

Pricing not listed

Pricing not listed — check our directory.

The rollback fixed the symptom, not the exposure

Any AI image generator with a satellite aesthetic can replicate this risk outside of Google Earth, which means removing the feature from one product does not close the workflow for bad actors.

Alternatives worth knowing

Genuine satellite imagery platforms like Maxar and Planet Labs maintain strict chain-of-custody documentation that AI-generated overlays cannot fake. If your work depends on verifiable geospatial evidence, the verification layer now has to live with you, not with the platform.

The line between real and generated satellite imagery is dissolving fast

This is exactly the kind of shift we track every week — who shipped what, what broke, and what it means for professionals who depend on these tools. We cover tools like this every Friday — subscribe here and we’ll send the best ones straight to you.